Lesson 10 of 14 · about 55 minutes · Updated November 15, 2026
Privacy Programs and Data Inventories
You'll be able to
- Describe the roles of a chief privacy officer, privacy counsel, a data protection officer and privacy champions, and compare reporting lines for the privacy office.
- Build and maintain a data inventory and records of processing using GDPR Article 30, OMB Circular A-130 and Minnesota Statutes 325M.18 as models.
- Identify when a privacy impact assessment, GDPR DPIA, state data protection assessment or California risk assessment is required, and what it must contain.
- Explain how to run consumer rights requests at scale, including verification limits, Global Privacy Control and California's 10 million consumer metrics rule.
- Apply data minimization, retention and tracking technology controls in light of the Maryland law, the GM, GoodRx and BetterHelp cases and the HHS tracking guidance ruling.
- Report privacy program performance to leadership and find the public records that show how an organization handles personal data.
Start here: the short version
- A privacy program is a machine that answers five questions on demand: what personal data do we hold, where is it, why do we have it, who receives it, and what have we promised the people it is about. Every privacy law, from the EU's General Data Protection Regulation to California's regulations to the federal E-Government Act, asks some version of those questions.123
- The data inventory comes first. Assessments, rights requests, retention, breach notice and vendor contracts all fail without one. Federal agencies, Minnesota businesses and anyone subject to the GDPR's Article 30 must keep one in writing.451
- Written risk assessments are now the norm, not the exception. By one law firm's comparison chart, every broad state consumer privacy law except Utah's and Iowa's requires documented assessments for high-risk processing, and California's regulations, in force since Jan. 1, 2026, require them before risky processing starts, with summaries due to the state by April 1, 2028.67
- Regulators now punish collecting too much, not only losing it. California's $12.75 million settlement with General Motors on May 8, 2026 was its first built on data minimization and purpose limitation, and Maryland's law, applying to processing from April 1, 2026, limits collection to what is "reasonably necessary and proportionate."89
- Tracking pixels are the most common way large organizations break their own privacy promises. The FTC's GoodRx and BetterHelp orders (2023), the HHS tracking guidance and the court ruling that cut it back (2024), and California's 2025 and 2026 cases all turned on code that marketing added to a website.10111213
- Making rights hard to use is itself a violation. California fined Honda $632,500 in 2025 and Ford $375,703 in 2026 partly for demanding identity checks before honoring opt-outs.1413
- The NIST Privacy Framework is the free organizing tool. Version 1.0 (Jan. 16, 2020) is still the current final version; Version 1.1, aligned with Cybersecurity Framework 2.0, was still an initial public draft when this lesson was checked.1516
Rules and standards described: NIST Privacy Framework Version 1.0 (NIST CSWP 01162020, Jan. 16, 2020) and the Version 1.1 initial public draft (NIST CSWP 40 ipd, April 14, 2025); Regulation (EU) 2016/679 (GDPR) Articles 5, 25, 30, 35 to 39, as in force; E-Government Act of 2002 Section 208 and OMB Memorandum M-03-22 (Sept. 26, 2003); OMB Circular A-130 (July 2016); California Consumer Privacy Act (Civil Code 1798.100 and following) and the California Privacy Protection Agency regulations on risk assessments, cybersecurity audits and automated decisionmaking technology, effective Jan. 1, 2026, and 11 CCR 7102 as amended March 29, 2023; Maryland Online Data Privacy Act (effective Oct. 1, 2025); Minnesota Statutes 325M.18 (2026 edition); the amended COPPA Rule, 16 CFR Part 312 (90 FR 16918, April 22, 2025); Idaho Technology Authority policy P4150 and guideline G502; all read Nov. 15, 2026. Privacy law changes several times a year at the state level; check the regulator's current text before relying on any point.
Before you change anything. Plan changes to production systems, test them first, keep a way back, and follow your change management process. For clinical, plant-floor or other safety systems, involve the vendor, your safety lead and qualified engineers (see modules 109 and 111).
A regional health system with 14,000 employees decides to launch a new patient app. Marketing wants analytics on the app and the public website. The data science team wants to train a model that predicts which patients will miss appointments. A vendor offers to run the app's chat feature. The app will also reach patients in California, and the health system's insurance arm sells policies in Maryland and Minnesota. Before a line of code ships, someone in this organization has to be able to say what personal data the app will collect, which laws apply to each piece, whether a written assessment is required and by whom, what the privacy notice must say, what the vendor contract must contain, how long the data will be kept, and how a patient who asks to see or delete their data will be answered within the legal deadline.
In a 20-person firm, the owner and an IT contractor can answer those questions from memory. In a large organization nobody can. The data is spread across hundreds of systems and dozens of vendors, the laws differ by state and by sector, and the people who add new data flows (marketing, product teams, researchers, human resources) are not the people who know the rules. A privacy program is the set of people, records and routines that lets a large organization answer those questions reliably, prove that it did, and change course when the law changes.
This lesson is about running that program. The legal map, with every major federal and state privacy law, the GDPR's reach into U.S. companies and a starter privacy notice and inventory worksheet, is in Module 108, Privacy Law Overview, Resource and Policy Library; this lesson does not repeat it. The small-business version of the privacy program is in Module 73, Privacy Program, AI Governance and Board Reporting, with data mapping basics in Risk Assessment and Data Mapping. The full course on U.S. privacy statutes is Module 69, Privacy and Data Security Laws. Inside this module, the policy machinery that privacy policies share with security policies is in lesson 4, privacy metrics for the board in lesson 3, privacy diligence on acquired companies in lesson 11, and the governance of artificial intelligence, which leans on the same inventories and assessments, in lesson 13.
What a privacy program is, and how it differs from security
Security protects data from people who should not have it. Privacy governs what the organization itself does with data about people: whether it should collect the data at all, what it may use it for, who it may share it with, how long it keeps it, and what rights the person keeps over it. A perfectly secure system can still violate privacy law. GoodRx, a prescription discount service, was not hacked; the FTC's 2023 case was about the company itself sending health information to advertising platforms in breach of its own promises.10 General Motors was not hacked either; California's 2026 case was about selling driving data for purposes drivers had not been told about.8
The two functions share a great deal: the same data inventory, the same vendor contracts, the same incident response plan, the same classification scheme. Large organizations that run them as strangers end up with two inventories that disagree and two vendor questionnaires that ask the same questions. Organizations that merge them completely often find that privacy questions (should we do this at all?) get lost under security questions (is it encrypted?). The usual answer is two leaders who share tools and records and meet on a schedule.
| Question | Security program asks | Privacy program asks |
|---|---|---|
| Collection | Can we protect this data once we have it? | Do we need it at all? Did we tell people? Is it sensitive data that needs consent or is barred? |
| Use | Who inside can reach it, and is access logged? | Is this use compatible with the purpose we disclosed? Does it trigger an assessment? |
| Sharing | Is the vendor secure? Is the transfer encrypted? | Is this a "sale" or "sharing" under state law? Does the contract limit the vendor's use? Is a cross-border transfer allowed? |
| Retention | Are backups protected and restorable? | Why do we still have it? When must it be deleted? |
| Individuals | Is the account protected from takeover? | Can the person see, correct, delete or opt out, and can we prove we answered on time? |
| Failure | Breach of security, ransomware, insider theft | Unlawful sharing, over-collection, broken promises, ignored opt-outs, as well as breaches |
The rules that require a program, not just compliance
Many privacy laws impose duties one at a time. A growing number require the organization to have a program: people, written policies and records. The main sources for a large U.S. organization:
- Federal agencies. OMB Circular A-130, revised in July 2016, requires each agency to establish and maintain an agency-wide privacy program, to designate a Senior Agency Official for Privacy (SAOP) with agency-wide responsibility, and to maintain an inventory of the information systems that create, collect, use or share personally identifiable information.4
- Minnesota. The Minnesota Consumer Data Privacy Act requires covered controllers to "document and maintain a description of the policies and procedures" they use to comply, including, where applicable, contact information for a "chief privacy officer or other individual with primary responsibility," and to keep "an inventory of the data that must be managed" as part of their security practices.5 That makes a written data inventory an express statutory duty in Minnesota.
- FTC and state orders. Almost every FTC privacy order since the early 2010s, and many state settlements, require the company to build and keep a privacy program for 20 years, with independent assessments. BetterHelp's 2023 order requires a standing privacy program; California's 2026 GM settlement requires a privacy program to "assess, mitigate and document privacy risks," with privacy assessments and reports.118 A program built before an order costs far less than one built under it.
- The GDPR. An organization reached by the GDPR must be able to demonstrate compliance (the "accountability" principle in Article 5(2)), keep records of processing (Article 30), assess high-risk processing (Article 35), build data protection into systems by design and by default (Article 25), and in some cases appoint a data protection officer (Articles 37 to 39).1
- Sector rules. HIPAA requires covered entities to designate a privacy official (45 CFR 164.530(a)), covered in Module 109, The Privacy Rule and Part 2. The amended COPPA Rule requires a written children's personal information security program and a written retention policy (lesson 108-12 covers the details).17
- Idaho state agencies. The Idaho Technology Authority (ITA) publishes policy P4150, Privacy Policy, and guideline G502, Privacy Impact Assessment, for state agencies under its authority.18 Idaho has no broad consumer privacy law as of the 2026 session; Idaho businesses feel these duties through the laws of the states where their customers live.
The privacy office
There is no single right shape for a privacy office, but large organizations converge on a small central team that sets the rules and keeps the records, a legal function that interprets the law, and a network of people inside each business unit who do the daily work.
The chief privacy officer
The chief privacy officer (CPO) owns the privacy program: its policies, its inventory, its assessment process, its rights-request operation, its training and its reporting to the executive team and board. Some organizations give the job a different title (privacy director, head of privacy, data protection officer, Senior Agency Official for Privacy in a federal agency). What matters is the authority in the written charter: whether the CPO can stop a launch that has not been assessed, whether the CPO reports to someone with the standing to back that decision, and whether the CPO has a direct line to a board committee.
A CPO's yearly work usually includes:
- Keeping the privacy policy set (top-level policy, standards for notices, consent, retention, rights requests, tracking technologies, vendor terms, cross-border transfers) current and approved, using the policy process in lesson 4.
- Owning the data inventory and records of processing with the security team and data owners.
- Running the assessment process: intake, triage, review, sign-off, retention of the record.
- Running or overseeing the consumer, patient, employee and customer rights request operation and its metrics.
- Reviewing public privacy notices, cookie banners and consent flows before they change.
- Advising on incidents: whether a security incident involves personal data, which notice laws apply (lesson 108-11, Breach Notification), and what the organization has promised.
- Tracking new laws and regulations and turning them into requirements with owners and dates.
- Training: everyone yearly, and deeper training for marketing, product, data science, HR and customer service.
- Reporting to leadership: open assessments, request volumes and timeliness, incidents, regulator contacts, and new legal risks.
Privacy counsel and the general counsel's office
In most large companies, privacy lawyers sit inside the general counsel's office. They interpret statutes and regulations, draft and negotiate data processing agreements, answer regulator inquiries, manage litigation such as wiretap and pixel class actions, and decide when work should be done at the direction of counsel to preserve privilege. The CPO may be one of those lawyers, or may be a non-lawyer who works alongside them. Each model has costs. A CPO inside the legal department carries the department's authority and privilege but may be seen as the person who says no. A CPO in compliance or risk is closer to operations but must route legal questions through the general counsel's office. The California regulations make one choice visible: the people who contributed to a risk assessment must be named in it, except counsel giving legal advice, which means the assessment itself is a business record and the privileged analysis must be kept separate.19
The data protection officer
The GDPR requires a data protection officer (DPO) for public authorities, for organizations whose core activities involve large-scale regular and systematic monitoring of people, and for those whose core activities involve large-scale processing of special categories of data such as health data (Article 37). The DPO must be involved in all data protection issues "properly and in a timely manner," must not be dismissed or penalized for doing the job, must report to "the highest management level," and must not hold a conflicting role (Article 38).1 European regulators have read that last point to bar heads of legal, IT or marketing from also serving as DPO in many cases. A U.S. multinational with an EU subsidiary often appoints a DPO in Europe who reports to the global CPO on operations and to the board on independence.
Privacy champions and the federated model
No central team of ten can review every marketing campaign, product change and spreadsheet in a 50,000-person company. Large programs use a federated model: each business unit names one or more privacy champions (sometimes called privacy liaisons or data stewards), trained to spot the triggers that need the central team, to keep their unit's part of the inventory current, and to route rights requests and incidents. Champions are usually part-time, so the program must give them short checklists, a clear escalation path, and recognition in their performance reviews; otherwise the role becomes a name on a list.
The federated model has a known failure: champions who are also the people pushing a launch. The fix is the same as for security exceptions in lesson 4: champions triage and collect facts, but only the central office (or a designated reviewer outside the business unit) signs off on a high-risk assessment.
Government privacy offices
In the federal government, Executive Order 13719 (Feb. 9, 2016) required agency heads to designate a Senior Agency Official for Privacy, and OMB Circular A-130 assigns that official agency-wide responsibility, including review of the information technology budget request for privacy requirements and costs.4 Federal privacy offices also run Privacy Act work: system of records notices published in the Federal Register, which tell the public what records an agency keeps about people and how they are used. Many state governments have a state chief privacy officer; Idaho's privacy rules for agencies come through the Idaho Technology Authority's policies rather than a statute creating a privacy office.18 For counties, cities, school districts and hospital districts, privacy work usually falls to the clerk, records custodian, compliance officer or IT director, which is one more reason to use a short, standard assessment form.
Where the privacy office sits
| Reporting line | Strengths | Risks | Common in |
|---|---|---|---|
| General counsel | Legal authority and privilege; close to litigation and regulators | Seen as a legal checkpoint; operations work may be underfunded | Technology, consumer brands, health systems |
| Chief compliance or chief risk officer | Fits with other compliance programs, testing and monitoring | May treat privacy as checklist compliance; less product influence | Banks, insurers, regulated finance |
| CISO or CIO | Shares tools and inventories with security; close to systems | Privacy questions reduced to security questions; conflict when IT wants a project | Mid-size organizations, some public agencies |
| Chief data officer | Close to data governance, quality and analytics | The office that wants to use data also judges its use | Data-heavy enterprises, some state governments |
| Chief executive (direct) | Maximum authority | Rare; depends on the executive's attention | Companies whose product is personal data, after major enforcement |
Whatever the line, the CPO should have documented access to the board or a board committee, as the CISO does under the rules covered in Module 108, Governance: Boards, Officers and Disclosure, and the board's risk reporting should include privacy risk alongside cyber risk.
The NIST Privacy Framework
The National Institute of Standards and Technology published the NIST Privacy Framework Version 1.0 on Jan. 16, 2020, as a free, voluntary tool for managing privacy risk. It is built like the Cybersecurity Framework: a Core of functions, categories and subcategories (outcomes to achieve, not controls to buy), Profiles that describe where an organization is and where it wants to be, and Implementation Tiers that describe how mature its risk management is.20
The five functions
| Function (Version 1.0) | What it covers | Program work it maps to |
|---|---|---|
| Identify-P | Understanding the data processing, the people affected and the risks: inventory and mapping, business environment, risk assessment, risk management for vendors and partners in the data processing chain | Data inventory, records of processing, data maps, assessments, vendor list |
| Govern-P | Policies, processes and procedures; roles; legal and regulatory requirements; awareness and training; monitoring and review | Privacy office charter, policy set, training, regulatory tracking, board reporting |
| Control-P | Managing data at a granular level: data processing policies, management of data (including retention and disposal), disassociated processing (de-identification and similar techniques) | Retention schedules, minimization, de-identification standards, rights request fulfillment |
| Communicate-P | Making sure the organization and individuals understand how data is processed: transparency and processing awareness | Privacy notices, just-in-time notices, consent records, regulator and public communication |
| Protect-P | Data protection safeguards: identity management, data security, maintenance, protective technology | Shared with the security program; often satisfied by Cybersecurity Framework work |
Version 1.0 uses the hyphen-P labels so that privacy and cybersecurity outcomes can sit side by side without confusion. The Identify-P category on inventory and mapping is the one most organizations start with, because it is the foundation for every other function.20
Version 1.1: still a draft
On April 14, 2025, NIST released an initial public draft of Privacy Framework 1.1 (NIST CSWP 40 ipd), with comments taken until June 13, 2025. The draft restructures the Core to align with Cybersecurity Framework 2.0 (including its emphasis on governance and risk management strategy), adds a section on how artificial intelligence creates privacy risks such as data reconstruction and inferences about people, and moves the user guidance into a separate online resource.16 As of NIST's Privacy Framework page, last modified June 24, 2026 and read Nov. 15, 2026, Version 1.0 remained the current final version and 1.1 was still labeled an initial public draft.15 A program can use the 1.1 draft's structure now to line up with its Cybersecurity Framework 2.0 profile (lesson 108-03, The Standards Map), but should cite Version 1.0 as the standard it follows until NIST finalizes the update.
Using the framework in a large organization
- Write a Current Profile. For each subcategory, record whether the outcome is achieved, partly achieved or not achieved, with the evidence (a policy, a record, a system report). Do it by business unit if the units differ.
- Write a Target Profile. Choose which outcomes matter most given the laws that apply, the data held and the organization's risk appetite. Not every subcategory needs the same depth.
- Turn the gap into a plan. Each gap becomes a project with an owner, budget and date, tracked in the same register as security risks (lesson 108-02, Risk Management in Plain Terms).
- Map the laws to the profile. NIST publishes crosswalks from the Privacy Framework to other frameworks and laws; a unified control set lets one assessment answer many laws, as described in lesson 4.
- Report the tier honestly. Tiers 1 to 4 (Partial, Risk Informed, Repeatable, Adaptive) describe how risk decisions are made, not how many controls exist. A program that does assessments only when someone remembers is Tier 1 or 2, whatever its policy binder says.
Data inventories and records of processing
A data inventory is a list of the personal data an organization holds, organized so that people can use it: by system, by business process, or both. A record of processing activities (often shortened to ROPA) is the GDPR's version: a written record, organized by processing activity (payroll, marketing email, patient scheduling, fraud screening), of what is done with personal data and why. In practice large organizations keep one inventory that can produce both views.
Why the inventory comes first
Every other part of the program depends on it:
- Assessments need to know what data a new project touches and what it will be combined with.
- Rights requests cannot be answered on time if nobody knows which of 400 systems holds a person's data.
- Privacy notices must list categories of data, purposes, sources and recipients; California's law requires those categories by name.21 A notice written without an inventory is a guess, and a wrong guess is a deceptive statement under Section 5 of the FTC Act.22
- Breach notification depends on knowing, within hours, whose data was in the affected system and which states they live in.
- Retention and deletion need a list of what exists before anything can be deleted on schedule.
- National security rules. The Justice Department's Data Security Program, 28 CFR Part 202, effective April 8, 2025, restricts transfers of bulk U.S. sensitive personal data (such as precise geolocation, health, financial and biometric data above set volume thresholds) to countries of concern and covered persons. An organization cannot tell whether it crosses a bulk threshold without an inventory that counts records by category.23
- Security. Classification, encryption, access reviews and backup priorities all start from knowing where sensitive data lives (lesson 108-07, Data Protection, Encryption, Retention and Backups).
GDPR Article 30 as a model
Even an organization the GDPR does not reach can use Article 30 as a checklist, because it is the most widely used template in the world for what an inventory should contain. Article 30(1) requires a controller's record to contain:1
- the name and contact details of the controller (and any joint controller, representative and data protection officer);
- the purposes of the processing;
- a description of the categories of data subjects and the categories of personal data;
- the categories of recipients to whom the data have been or will be disclosed, including recipients in other countries;
- transfers to a third country or international organization, and the safeguards used where required;
- where possible, the time limits for erasure of the different categories of data; and
- where possible, a general description of the technical and organizational security measures.
Article 30(2) requires processors (service providers acting on a controller's instructions) to keep a parallel record of the processing they do for each controller. The record must be in writing, including electronic form, and must be made available to the supervisory authority on request (Article 30(3) and (4)). Article 30(5) exempts organizations with fewer than 250 employees unless the processing is likely to result in a risk to people's rights, is not occasional, or includes special categories of data or criminal records data; since nearly every employer processes payroll data regularly, the exemption is narrower than it looks.1 The European Union is negotiating a package of simplifications to its data and cyber rules; the Council of the EU agreed its position on Oct. 21, 2026, including a single EU-wide list of high-risk processing and common templates, but the changes are not law and the final text may differ.24
U.S. state laws add fields Article 30 does not have. A large organization's inventory usually also records the states whose residents' data is held, whether each data element is "sensitive" under each state's definition, whether a flow counts as a "sale," "sharing" or "targeted advertising," the legal basis or exemption relied on (for example, HIPAA protected health information or Gramm-Leach-Bliley data exempt from a state law), and the approximate number of people, which drives both state law thresholds and the Justice Department's bulk thresholds.
Laws that require an inventory in writing
| Source | Who it reaches | What it requires |
|---|---|---|
| GDPR Article 30 | Controllers and processors subject to the GDPR | Written records of processing with the fields listed above, available to regulators1 |
| OMB Circular A-130 (2016) | Federal executive agencies | Inventory of information systems that create, collect, use, process, store, maintain, disseminate, disclose or dispose of personally identifiable information4 |
| Minnesota Statutes 325M.18 | Controllers covered by the Minnesota Consumer Data Privacy Act | "An inventory of the data that must be managed," plus documented privacy policies and procedures5 |
| California CCPA and regulations | Businesses meeting the CCPA thresholds | No inventory by that name, but notices at collection, privacy policies, risk assessments and request metrics all require inventory-level facts2119 |
| Idaho Technology Authority P4130 and G505 | Idaho state agencies under ITA policy | Classification of information systems and data, which requires knowing what each system holds18 |
Building the inventory at scale
Large organizations build the inventory from three directions at once and reconcile the results, because each direction misses things the others catch.
- Top down, by business process. Interview each process owner with a short questionnaire: what personal data do you collect, from whom, why, where is it stored, who receives it, how long do you keep it. Start with the processes that hold the most sensitive data or the most people: customer accounts, patient records, payroll and benefits, marketing, call centers, fraud and collections.
- Bottom up, by system. Use the IT asset inventory and the configuration management database as the list of systems; for each, record the data elements, the owner and the vendor. Automated data discovery tools can scan databases, file shares and cloud storage for patterns that look like personal data (identification numbers, account numbers, health terms) and are good at finding copies nobody remembered. They produce false positives and miss data in unstructured formats, so their results feed the inventory; they do not replace it.
- Outside in, by vendor. The accounts payable list and the contract repository show every vendor paid in the last year. Any vendor that receives personal data belongs in the inventory as a recipient, and the contract must match (lesson 108-08, Vendors and Supply Chain). The website's tag list is a vendor list too, as the tracking section below explains.
A realistic first pass at a large organization takes three to nine months and produces an inventory that is perhaps 80 percent right. That is acceptable if the program says so, keeps a list of known gaps, and has a way to keep improving.
Keeping it current
An inventory decays from the day it is finished. The organizations that keep theirs useful connect it to processes that already happen:
- Procurement: no purchase order for software or a service that touches personal data without an inventory entry and, where triggered, an assessment.
- Change management: a new data element, new recipient or new purpose in a change ticket updates the inventory.
- Architecture review and the security review of new systems share one intake form with privacy.
- Annual attestation: each data owner confirms or corrects their entries once a year, with the confirmation recorded.
- Incidents and rights requests: when a breach investigation or a deletion request finds data that was not in the inventory, the gap is logged and fixed.
- Mergers: an acquired company's inventory, or the lack of one, is a diligence item (lesson 11, Security in Mergers and Acquisitions).
Some organizations keep the inventory in a spreadsheet; most large ones move to a privacy management platform or the governance, risk and compliance tool described in lesson 4. The tool matters less than ownership: each entry needs a named owner who answers when it is wrong.

Data mapping
An inventory says what data exists; a data map shows how it moves: from the person, through the organization's systems, to vendors, partners, regulators and other countries, and finally to deletion. Maps are what make privacy risk visible to people who are not privacy specialists. A board member who would skim a 900-row spreadsheet understands at once a picture in which a patient's appointment data flows from the hospital's website to an advertising platform.
What a data map shows
A useful map, for each important process, shows the sources of data (the person, a third party, a device), each system that stores or processes it, each internal team that uses it and for what purpose, each external recipient and the legal character of the transfer (service provider, third party, sale, sharing, regulator), cross-border flows, and the retention point where the data is deleted or de-identified. Lines should be labeled with the data categories that travel on them, because "customer data" hides the difference between an email address and a diagnosis.
From map to decisions
The point of the map is to drive decisions. Each flow should be checked against four questions: is it disclosed in the notice; is there a contract with the right terms for that recipient's role; is it a sale, sharing or targeted advertising flow that requires an opt-out or consent in some states; and does it cross a border or a national security threshold. Flows that fail any of these become remediation items with owners. The map is also the first thing to pull when a regulator, a plaintiff or a reporter asks where data went.
Privacy impact assessments and data protection assessments
A privacy impact assessment (PIA) is a written review, done before a new system, project, vendor or use of data begins, of what personal data is involved, the risks to the people it describes, and the steps that will reduce those risks. Different laws use different names: data protection impact assessment (DPIA) under the GDPR, data protection assessment under most U.S. state laws, data privacy and protection assessment in Minnesota, risk assessment under California's regulations. The core idea is the same: think before you process, write it down, and have someone with authority decide.
The federal PIA under Section 208
The idea came from government. Section 208 of the E-Government Act of 2002 requires federal agencies to conduct a PIA before developing or buying information technology that collects, maintains or disseminates information in identifiable form, and before starting a new collection of identifiable information from 10 or more people (other than federal employees). The PIA must be reviewed by the agency's chief information officer or an equivalent official and, if practicable, made public. OMB Memorandum M-03-22, dated Sept. 26, 2003, sets out what the PIA must cover: what information is collected, why, its intended use, with whom it will be shared, what notice and opportunities for consent individuals get, how it will be secured, and whether a Privacy Act system of records is created.2 That is why federal agency websites carry long lists of published PIAs, which are useful public records (see the reporting section below). The Idaho Technology Authority's guideline G502 adapts the same approach for Idaho state agencies.18
GDPR Article 35 DPIAs
Under Article 35, a controller must carry out a DPIA before processing that is "likely to result in a high risk to the rights and freedoms of natural persons," taking into account the nature, scope, context and purposes of the processing, and in particular when using new technologies. Article 35(3) names three cases that always require one: systematic and extensive evaluation of people based on automated processing, including profiling, that produces legal or similarly significant effects; large-scale processing of special categories of data or criminal records data; and systematic monitoring of a publicly accessible area on a large scale. Each national supervisory authority publishes a list of further processing that requires a DPIA (Article 35(4)).1
Article 35(7) sets the minimum content: a systematic description of the processing and its purposes; an assessment of necessity and proportionality; an assessment of the risks to people; and the measures to address those risks, including safeguards and security measures. The controller must seek the advice of its data protection officer, and if the DPIA shows a high risk that the controller cannot reduce, it must consult the supervisory authority before starting (Article 36).1 The European data protection authorities' guidelines (WP248 rev.01, as last revised and adopted Oct. 4, 2017, later endorsed by the European Data Protection Board) list nine criteria that point to high risk, such as sensitive data, large scale, vulnerable people, combining data sets and innovative technology, and suggest that processing meeting two or more criteria will usually need a DPIA.25
State law data protection assessments
The U.S. state consumer privacy laws borrowed the DPIA. By one law firm's comparison chart, every state consumer privacy law it tracks requires documented data protection assessments except Iowa's and Utah's.6 The typical triggers are processing for targeted advertising, selling personal data, processing sensitive data, and profiling that presents a reasonably foreseeable risk of unfair or deceptive treatment, financial, physical or reputational injury, intrusion on seclusion, or other substantial injury. Minnesota's version, for example, requires a controller to "conduct and document a data privacy and protection assessment" for targeted advertising, sale of personal data, processing of sensitive data, processing that presents "a heightened risk of harm to consumers," and certain profiling.5 Most state laws let the attorney general demand the assessment, keep it confidential and exempt from public records laws, and accept one assessment for a set of similar processing operations, or an assessment done for another law if it is reasonably comparable in scope and effect. MultiState reported in October 2026 that Delaware's 2026 amendments will require more detailed assessments of consequential profiling, covering the data used, outputs, risks, performance measures and limitations.26
The count of states keeps moving. The IAPP's tracker, updated June 29, 2026, listed 23 signed state laws without Florida's narrow law; MultiState, writing Oct. 28, 2026, counted 24 including Florida after Alabama, Louisiana, Oklahoma and Vermont enacted laws in 2026.2726 This lesson did not confirm the assessment provisions of each 2026 law; check the enacted text before relying on the count.
California's risk assessment regulations
California's rules are the most detailed in the country. The California Privacy Protection Agency (which now calls itself CalPrivacy) adopted regulations on risk assessments, cybersecurity audits and automated decisionmaking technology (ADMT) on July 24, 2025; they were approved by the Office of Administrative Law and filed Sept. 22, 2025, and took effect Jan. 1, 2026.37 For risk assessments, as summarized by Morgan Lewis in July 2026:19
- Triggers. Six kinds of processing present "significant risk": selling or sharing personal information; processing sensitive personal information (with an exception for certain employment-context processing); using ADMT for a significant decision about a consumer; automated processing to infer or extrapolate traits in employment or educational settings; automated processing based on a person's presence at a sensitive location; and processing to train ADMT or facial recognition, emotion recognition or identity verification technology.
- Content. The specific purpose; the categories of personal information, including the minimum necessary; operational elements (how it is collected, used, disclosed and retained, the approximate number of consumers, recipients, and for ADMT its logic and how outputs are used); the benefits; the negative impacts on consumers' privacy and their sources; the safeguards; whether the business will proceed; the names of contributors (other than counsel giving legal advice); and the names, positions and dates of the reviewers and approvers, at least one of whom has authority to decide whether the processing will start.
- Timing. Before starting covered processing; for processing already under way before 2026, by Dec. 31, 2027; reviewed at least every three years; and updated within 45 calendar days of a material change.
- Retention. For as long as the processing continues or five years after completion, whichever is later.
- Submission. Businesses must begin compliance Jan. 1, 2026 and, by April 1, 2028, submit to the agency an attestation that required assessments were completed and a summary of the assessment information for assessments done in 2026 and 2027.7
The approval requirement changes how large organizations run assessments. A privacy analyst can no longer close an assessment alone; the record must show a named person with authority to go or not go, which pulls business leaders into the decision and gives the privacy office a written basis for escalating.
Cybersecurity audits and automated decisions
The same California package requires an annual independent cybersecurity audit for businesses whose processing presents significant risk to consumers' security (as defined in the regulations), with certifications of completion due to the agency on a schedule tied to annual gross revenue: by April 1, 2028 for businesses with more than $100 million; April 1, 2029 for $50 million to $100 million; and April 1, 2030 for less than $50 million. Businesses using ADMT to make significant decisions about consumers (such as decisions on lending, housing, employment, education or health care) must comply with notice, opt-out and access requirements beginning Jan. 1, 2027.7 Audit readiness belongs with the internal audit and assurance work in lesson 8; ADMT governance belongs with the AI program in lesson 13. For a program leader, the practical point is that one intake process should catch all three: a new use of AI on consumer data may need a risk assessment, ADMT notices and a check against the audit scope at the same time.
Running assessments at scale
A large organization may screen hundreds of projects a year and complete full assessments on dozens. The programs that keep up do several things:
- One intake, many laws. A single short screening form, built into procurement, architecture review and project approval, asks the trigger questions for every applicable law at once. The answers route the project to "no assessment," "short form" or "full assessment."
- Templates by type. Separate templates for vendor onboarding, new marketing technology, AI and analytics, HR systems and research save time and get better answers than one generic form.
- Clear decision rights. The template names who approves at each risk level: the privacy office for medium risk, a business executive with authority to proceed (and the CPO's concurrence) for high risk, and an executive committee for the highest. California's regulations now require the approver's name and position on the record.19
- Service levels. Publish how long reviews take (for example, five business days for a short form, 20 for a full assessment) and measure it. Slow reviews push teams to skip the process.
- Conditions that are tracked. Most assessments end with "proceed, if": shorten retention, drop a data field, add a contract term, turn off a tracking tag. Those conditions go into a tracker with owners and dates, and the privacy office checks that they were done before or soon after launch.
- Reuse. A well-written assessment for a vendor platform can cover later projects that use the same platform the same way, which the state laws generally allow when processing is similar.
- A defensible archive. Keep every version, with dates and approvers, for the longest retention period that applies (five years after completion under California's rules). Assume that a regulator, plaintiff or public records requester will read it.
Consumer rights requests at scale
State laws, the GDPR and sector laws give people rights to know what data an organization holds about them, to get a copy, to correct it, to delete it, to opt out of sale, sharing, targeted advertising and some profiling, and (in California) to limit use of sensitive personal information. A national retailer, bank or health system may receive thousands of these requests a month, through web forms, call centers, email, store staff, browser signals and authorized agents. A small company can answer them by hand. A large one needs a pipeline.
The request pipeline
- Intake. Offer the methods the laws require (California requires at least two methods for most requests, and businesses that operate online must offer a way to submit online) and treat any request that reasonably reaches the organization as received, including one sent to the wrong department.21
- Classify. Which right, which law (by the person's state or country), which deadline. California and most states allow 45 days for access, deletion and correction requests, extendable once by 45 days with notice; the GDPR allows one month, extendable by two months in some cases.211
- Verify, in proportion. Access and deletion requests need identity verification matched to the sensitivity of the data. Opt-out requests do not: California's regulations bar requiring verification for opt-outs, and the agency's cases against Honda (2025) and Ford (2026) both turned on adding verification steps to opt-outs.1413
- Find the data. This is where the inventory pays off. The pipeline queries each system that holds data about the person, or sends tasks to the owners of systems that cannot be queried automatically, and to service providers who must help.
- Apply exceptions and act. Legal holds, records retention duties, fraud prevention, HIPAA or Gramm-Leach-Bliley exemptions, and other lawful reasons to keep data are applied item by item, and the response explains them.
- Respond and record. Send the answer through a secure channel, record what was done and when, and keep the record for the period the law requires (California's regulations require 24 months of request records).3
- Honor signals automatically. Browser-based opt-out preference signals, such as the Global Privacy Control, must be honored as valid opt-out requests in California and in several other states that have adopted the same rule. Sephora's 2022 settlement with the California attorney general ($1.2 million) was the first public case on the point.28
Request metrics and the 10 million rule
California requires a business that knows or reasonably should know that it buys, receives for commercial purposes, sells, shares or otherwise makes available the personal information of 10 million or more consumers in a calendar year to compile metrics for the previous calendar year and disclose them by July 1 of every year in its privacy policy or on a linked page. The metrics are, for each request type (delete, correct, know, opt out of sale or sharing, and limit), the number received, complied with in whole or in part, and denied, plus the median or mean number of days to respond substantively. A business may report all individuals' requests rather than only Californians' if it says so.29 The rule, now 11 CCR 7102 (renumbered in 2022 from the original section 999.317(g) of the attorney general's 2020 regulations and amended March 29, 2023), makes a large company's rights operation partly public, which is useful to reporters and to competitors comparing response times.
Internal metrics should go further: requests by type, channel and state; percent closed within the deadline; average and longest time to close; verification failure rate; percent of opt-out signals honored automatically; number of systems that still require manual searching; and complaints or regulator inquiries about request handling. These belong in the board dashboard described in lesson 3.
Data minimization and retention
Data minimization means collecting, using and keeping only the personal data needed for a stated purpose. It was long treated as an aspiration in the United States. That changed in 2025 and 2026.
Maryland
The Maryland Online Data Privacy Act took effect Oct. 1, 2025 and applies to processing activities on or after April 1, 2026. It reaches businesses that control or process the personal data of at least 35,000 Maryland residents (or 10,000 if more than 20 percent of gross revenue comes from selling personal data). It limits collection of personal data to what is "reasonably necessary and proportionate to provide or maintain a specific product or service requested by the consumer," and collection or processing of sensitive data to what is "strictly necessary" for that purpose, a limit that consent does not override. It bans the sale of sensitive data outright, and bans sale and targeted advertising for consumers the business knows or should have known are under 18.9 The attorney general enforces it as an unfair or deceptive trade practice, with civil penalties of up to $10,000 per violation and $25,000 for repeat violations.30 For a large organization, Maryland's test is a design rule: every data field in a Maryland-facing form or app needs a documented reason tied to the service the person asked for.
California and the GM settlement
California's law has required since 2023 that collection, use, retention and sharing be "reasonably necessary and proportionate" to the purposes for which the data was collected or processed, or for another disclosed purpose compatible with the context of collection (Civil Code 1798.100(c)).21 On May 8, 2026, the California attorney general announced a $12.75 million settlement with General Motors, described as the largest CCPA penalty to date and the first CCPA settlement built on data minimization and purpose limitation. The state alleged that GM sold drivers' names, contact information, precise geolocation and driving behavior (speed, hard braking, rapid acceleration) to the data brokers LexisNexis Risk Solutions and Verisk without adequate notice or consent and for purposes unrelated to those disclosed. The terms, subject to court approval, bar GM from selling driving data to consumer reporting agencies for five years, require deletion of retained driving data within 180 days unless the consumer gives affirmative express consent, require GM to ask the two brokers to delete the data, and require a privacy program with privacy assessments and reporting.831
Retention schedules
Minimization in time is retention. A retention schedule lists each records category, the legal or business reason to keep it, the retention period, the event that starts the clock (account closed, last visit, end of employment), and the disposal method. California requires businesses to disclose how long they keep each category of personal information, or the criteria used to decide.21 The amended COPPA Rule requires a written data retention policy for children's personal information.17 The BetterHelp order requires a retention schedule.11 Records retention and legal holds at smaller scale are in Module 73, Records Retention, Legal Holds and Workplace Privacy. At large scale the hard part is not writing the schedule; it is making systems delete. Programs that succeed build deletion into new systems as a requirement, run deletion jobs on a schedule, report how much data was deleted, and treat old data found during a breach investigation as a program failure to fix. Deleting production data is a change to production systems: plan it, test it on copies, and have a way back (such as a short holding period) before running it at scale.

Privacy by design
Privacy by design means building privacy protections into systems and processes from the start, rather than adding them after launch. The GDPR made it a legal duty in Article 25: controllers must implement appropriate technical and organizational measures, "both at the time of the determination of the means for processing and at the time of the processing itself," and by default only personal data necessary for each specific purpose may be processed, which covers the amount collected, the extent of processing, the retention period and accessibility.1 U.S. law reaches the same place through minimization rules like Maryland's and California's and through FTC orders. In a large organization, privacy by design is not a slogan; it is a set of engineering requirements that show up in the same places as security requirements:
- Requirements templates. Product requirement documents include a privacy section: data elements and why each is needed, retention, who can see it, whether it is sold or shared, and which notices change.
- Defaults. New features ship with the most protective reasonable setting: optional data fields off, location sharing off until asked, sharing with partners off.
- Architecture patterns. Approved patterns for pseudonymization, de-identification, aggregation, consent storage, deletion and access logging, so teams do not invent their own.
- Design reviews. The privacy office joins architecture review for anything that triggers the screening form, the same way security does.
- Testing. Pre-release checks confirm that tracking tags, data flows and deletion jobs behave as the assessment said they would.
An international standard, ISO 31700-1:2023, describes privacy by design requirements for consumer goods and services; like other ISO standards it is sold, not free, and is described here rather than reproduced. ISO/IEC 27701, the privacy extension to ISO/IEC 27001, is covered with certifications in lesson 6.
Cookie, pixel and tracking technology governance
A tracking pixel is a small piece of code, supplied by an advertising or analytics company, that a website or app runs to report what visitors do: pages viewed, buttons clicked, forms submitted, often with device identifiers, IP addresses and sometimes email addresses. Software development kits (SDKs) do the same inside mobile apps. Cookies and similar technologies store identifiers in the browser. Marketing teams add these tools because they measure campaigns. Privacy programs fail on them because they are added outside procurement, change without notice when the vendor updates its code, and send data directly from the user's browser to the vendor, so the organization's own logs may never show the flow.
What the 2023 to 2026 cases show
- GoodRx (FTC, Feb. 1, 2023). The FTC's first enforcement action under its Health Breach Notification Rule. The FTC alleged that GoodRx shared users' prescription and health information with advertising platforms through tracking tools despite promising not to. GoodRx agreed to a $1.5 million civil penalty and a ban on sharing health information for advertising.10
- BetterHelp (FTC, final order July 14, 2023). The FTC alleged that the online counseling service disclosed email addresses, IP addresses and health questionnaire answers to Facebook, Snapchat, Criteo and Pinterest for advertising after promising to "only use or disclose personal health data for limited purposes." The order required $7.8 million for consumer refunds, banned sharing health data for advertising, required affirmative express consent before certain disclosures, required BetterHelp to direct third parties to delete the data, and required a privacy program and a retention schedule.11
- Hospitals and the HHS tracking guidance. HHS's Office for Civil Rights issued a bulletin on online tracking technologies for HIPAA covered entities in December 2022 and revised it in March 2024. It warned that tracking on patient portals and many public web pages could disclose protected health information to vendors without the authorizations or business associate agreements HIPAA requires. The American Hospital Association and others sued. On June 20, 2024, in American Hospital Association v. Becerra, No. 4:23-cv-1110, the U.S. District Court for the Northern District of Texas declared unlawful and vacated the part of the guidance (the "Proscribed Combination") that treated an IP address combined with a visit to an unauthenticated public web page about health conditions or providers as protected health information; the court found that portion "improperly creat[ed] substantive legal obligations." HHS withdrew its appeal on Aug. 29, 2024.32 HHS's page still carries the guidance with a notice of the court's order.12 The rest of the bulletin, including its treatment of tracking on authenticated patient portals, was not vacated, and hospitals still face wiretap and consumer class actions and state law claims over pixels. Module 109, The Privacy Rule and Part 2, covers the HIPAA side.
- Sephora (California attorney general, Aug. 24, 2022). $1.2 million for failing to disclose that it sold data through tracking on its site and failing to honor Global Privacy Control signals; the judgment required reporting to the attorney general on sales, service provider contracts and GPC handling.28
- Healthline (California attorney general, July 1, 2025). $1.55 million, then the largest CCPA settlement, over the health information website's sharing of data for advertising.33
- PlayOn Sports (CalPrivacy, March 2026). $1.10 million, the agency's first decision involving students and schools. The board found that the company's GoFan digital ticketing platform, used by about 1,400 California schools, made ticket holders click "agree" to tracking for targeted advertising before they could use their tickets, sent people to industry opt-out sites instead of offering its own opt-out, and ignored opt-out preference signals.34
- Ford (CalPrivacy, March 2026). $375,703 and an order to audit the tracking technologies on its website and honor opt-out preference signals including the Global Privacy Control, after the company required email verification before processing opt-outs.13
A tag governance process
Organizations that stay out of these cases run tracking technology the way they run any other vendor that receives personal data:
- Inventory every tag and SDK on every website and app, by scanning the live sites and reading the tag manager configuration, and record what data each sends and to whom. Repeat the scan monthly and after every major release.
- Classify each tag: strictly necessary, analytics, advertising, or other; and for each, whether it amounts to a sale, sharing or targeted advertising under state law, and whether it can see health, financial, children's or other sensitive data.
- Put the tag manager under change control. Only named people can publish tag changes; each new tag or new vendor goes through the screening form; marketing cannot add a pixel the privacy office has not seen.
- Keep sensitive pages clean. No third-party advertising tags on authenticated portals, health condition pages, financial account pages, children's sections, or forms that collect sensitive data, unless an assessment approved it with contracts and consent in place.
- Match the banner to reality. Test that the consent banner and opt-out choices actually stop the tags they claim to stop, and that Global Privacy Control signals are honored. The California regulators test this themselves.
- Contract for the role. A vendor that will use data only for the organization's purposes can be a service provider with the right contract; one that uses data for its own advertising is a third party, and the flow is a sale or sharing.
- Audit and report. Report tag counts, unapproved tags found and fixed, and consent test results to leadership.
Removing or changing tags on a live website can break measurement, payments or forms; plan the change, test it, and have a way back.
Children's and teens' privacy
The FTC's amended COPPA Rule, published April 22, 2025 (90 FR 16918), took effect June 23, 2025, with compliance required by April 22, 2026 for most provisions. For a large organization the program changes are the important part: a written children's personal information security program with a designated coordinator and annual risk assessments, a written retention policy that is disclosed and that keeps data no longer than reasonably necessary, and separate verifiable parental consent before disclosing children's data to third parties, including for targeted advertising.17 The rule's full content is in lesson 108-12. The largest COPPA penalty remains the $275 million Epic Games paid in December 2022.35 State laws now reach teens as well: Maryland bars sale and targeted advertising for consumers known or reasonably believed to be under 18,9 and several states have age-appropriate design or minors' online safety laws, some of them enjoined in court. The program implications are an age signal in the inventory (which systems hold data on minors), stricter defaults for those systems, and assessments for any service likely to be used by young people.
Employee privacy
Large employers hold some of the most sensitive data about the people they know best: health and disability information, background checks, Social Security numbers and bank accounts for payroll, performance and discipline records, location data from company phones and vehicles, and increasingly monitoring data from productivity, security and AI tools. California is the only state whose consumer privacy law fully covers employees and job applicants; the exemption for employee data ended Jan. 1, 2023, so California workers have rights to know, delete, correct and limit, and employers must give notice at collection.21 Most other state consumer privacy laws exclude data collected in an employment context. Even so, employee data is covered by HIPAA (for a company's health plan), the Americans with Disabilities Act's confidentiality rules for medical information, the Fair Credit Reporting Act for background checks, state biometric laws (Illinois, Texas and Washington), state laws on monitoring and social media passwords, and, for European staff, the GDPR.
The program duties are the same as for customers: the HR systems belong in the inventory, new monitoring tools go through the screening form (California's risk assessment triggers include automated inferences about people in employment settings), retention schedules cover personnel files, and employees get a clear notice. Monitoring for security purposes (data loss prevention, insider risk tools) needs special care: the security team's case for visibility and the employees' expectation of privacy both have to be written into the assessment and the policy. Module 73, Records Retention, Legal Holds and Workplace Privacy, and Module 108, People, Training, Phishing and Insider Risk, cover the basics.
Enforcement: what regulators have done
Privacy enforcement in the United States is spread across the FTC, HHS, state attorneys general, California's privacy agency and private class actions. The cases below are the ones that most shape how large programs operate.
| Date | Regulator and company | Amount | What it was about | Program lesson |
|---|---|---|---|---|
| Aug. 24, 2022 | California AG v. Sephora | $1.2 million | Undisclosed sale via tracking; ignored Global Privacy Control28 | Tags are sales; honor signals |
| Feb. 1, 2023 | FTC v. GoodRx | $1.5 million | Health data to advertisers; first Health Breach Notification Rule case10 | Promises must match practice |
| July 14, 2023 | FTC, BetterHelp order | $7.8 million | Mental health data to ad platforms11 | Program, retention schedule, consent |
| July 30, 2024 | Texas AG v. Meta | $1.4 billion | Facial recognition data without consent under Texas's biometric law36 | Biometrics need consent |
| Jan. 2025 | Texas AG v. Allstate and Arity (suit filed) | Pending | First suit under the Texas Data Privacy and Security Act; driving and location data3738 | SDK data collection is collection |
| March 12, 2025 | CalPrivacy, American Honda | $632,500 | Excess verification and friction in rights requests14 | Do not verify opt-outs |
| May 2025 | Texas AG and Google | $1.375 billion | Location tracking, Incognito disclosures, biometric identifiers37 | State AGs act at scale |
| July 1, 2025 | California AG v. Healthline | $1.55 million | Sharing data for advertising33 | Health-adjacent content is sensitive |
| Sept. 30, 2025 | CalPrivacy, Tractor Supply | $1.35 million | Notices and opt-out failures, including for job applicants39 | Employee and applicant data count |
| March 2026 | CalPrivacy, PlayOn Sports | $1.10 million | Forced tracking consent for school event tickets34 | Captive audiences, minors |
| March 2026 | CalPrivacy, Ford | $375,703 | Email verification before opt-outs13 | Friction is a violation |
| May 8, 2026 | California AG v. General Motors | $12.75 million | Sale of driving data; data minimization and purpose limitation8 | Collect and use only for disclosed purposes |
The California attorney general's enforcement page also lists 2025 and 2026 settlements with Sling TV and Dish Media Sales ($530,000, Oct. 30, 2025), an education software company ($3.25 million, Nov. 6, 2025), Jam City ($1.4 million, Nov. 21, 2025) and Disney ($2.75 million, Feb. 11, 2026).31 CalPrivacy has also brought a series of actions under the Delete Act against data brokers that failed to register or honor deletion requests.40

Reporting and measuring the privacy program
Boards and executive teams need privacy reporting as much as cyber reporting, and the two should use the same format (see lesson 3). A useful quarterly privacy report covers:
- Coverage: percent of systems and vendors in the inventory with a named owner and a current attestation; known gaps.
- Assessments: number screened, number of full assessments, average review time, open conditions past due, high-risk projects approved with residual risk and who approved them.
- Rights requests: volume by type, percent on time, longest open, opt-out signals honored.
- Tracking: tags on the sites, unapproved tags found, consent test results.
- Minimization: data deleted under the retention schedule; systems without working deletion.
- Incidents: privacy incidents (including misdirected mail and unauthorized disclosures, not only hacks), notices given, regulator contacts.
- Legal change: new laws and regulations, their effective dates, and readiness.
- Enforcement in the sector: cases against peers and whether the same facts exist here.
Internal audit should test the privacy program on a cycle, as it tests security (lesson 8), and the program's budget and staffing should be argued with the same method as security budgets (lesson 2). Cyber insurance applications increasingly ask about privacy practices and exclude or sublimit certain privacy claims, such as wrongful collection and pixel litigation; the privacy office should read the policy wording with the insurance buyer (lesson 9).
Sample: Data Inventory and Records of Processing Worksheet
Module 108 offers a starter inventory worksheet organized by data set. This version is built for a large organization: one row per processing activity, with the fields needed to produce a GDPR Article 30 record, state law notices and assessments, and request routing from the same entry. Most organizations keep it in a database or privacy platform rather than a document; the fields are what matter.
Sample only. Adapt it to your organization's risks, systems and laws, have qualified people review it, and do not treat it as a substitute for a risk assessment.
Follows: GDPR Article 30(1) and (2); OMB Circular A-130 (2016) PII inventory; Minnesota Statutes 325M.18 (2026); the notice and assessment content in the California Consumer Privacy Act and the CalPrivacy regulations effective Jan. 1, 2026; NIST Privacy Framework 1.0, Identify-P inventory and mapping category; 28 CFR Part 202 bulk data categories; Idaho Technology Authority P4130 and G505 for Idaho agencies.
Part A. Identification and ownership
| Field | What to record | Example entry |
|---|---|---|
| A1. Activity ID and name | Unique ID and plain name of the processing activity | [PA-0142] [Patient appointment reminders] |
| A2. Business unit and process owner | Name, title, contact; the person who answers when the entry is wrong | [Ambulatory Operations; Director, Patient Access] |
| A3. Privacy champion | Unit's privacy liaison | [Name] |
| A4. Controller or processor role | Are we deciding the purposes (controller or business) or acting for someone else (processor or service provider)? Joint controllers? | [Controller] |
| A5. Legal entities and countries | Which subsidiaries process; where | [Health System Inc.; United States only] |
Part B. People, data and purpose
| Field | What to record | Example entry |
|---|---|---|
| B1. Categories of people | Customers, patients, employees, applicants, minors, website visitors, others | [Adult patients; parents of minor patients] |
| B2. Approximate number of people and states or countries | Counts drive state thresholds, request planning and federal bulk data thresholds | [About 410,000; ID, OR, WA, UT, MT] |
| B3. Data elements | List each element; flag sensitive data under each applicable law | [Name, phone, email, appointment date, department (sensitive: health)] |
| B4. Sources | From the person, another organization, a device, public records, inferences | [Patient; scheduling system] |
| B5. Purpose(s) | Specific purposes, not "business operations" | [Remind patients of scheduled visits] |
| B6. Legal basis or exemption | GDPR basis if applicable; HIPAA treatment, payment or operations; GLBA; consent; state law exemption relied on | [HIPAA health care operations; state laws exempt PHI] |
| B7. Is necessity documented? | Why each element is reasonably necessary and proportionate (and strictly necessary for sensitive data in Maryland) | [Department needed to give location; diagnosis not collected] |
Part C. Systems, recipients and transfers
| Field | What to record | Example entry |
|---|---|---|
| C1. Systems and storage locations | Applications, databases, cloud services, file shares, backups | [Scheduling module of EHR; messaging vendor platform] |
| C2. Internal recipients | Teams with access and why | [Patient access staff; IT support] |
| C3. External recipients and their role | Service provider, processor, business associate, third party, regulator; contract ID | [Messaging vendor: business associate; BAA #2024-118] |
| C4. Sale, sharing or targeted advertising? | Under each state's definitions; opt-out or consent method if yes | [No] |
| C5. Cross-border transfers | Countries; mechanism (Data Privacy Framework, standard contractual clauses); countries of concern under 28 CFR Part 202 | [None] |
| C6. Tracking technologies involved | Tags or SDKs that receive data from this activity | [None permitted on reminder pages] |
Part D. Retention, security, assessments and rights
| Field | What to record | Example entry |
|---|---|---|
| D1. Retention period and trigger | From the retention schedule; disposal method | [Message logs 2 years after send; deleted by scheduled job] |
| D2. Security classification and key safeguards | Classification level; encryption, access control, logging | [Confidential; encrypted at rest; role-based access] |
| D3. Assessments | Screening result; PIA, DPIA, state or California assessment ID, date, approver, next review | [Screened 2026-03-02; full assessment PIA-0311 approved by [VP]; review by 2029-03] |
| D4. Notices | Which privacy notice covers it; notice at collection | [Notice of Privacy Practices v. [date]] |
| D5. Rights request routing | How access, deletion, correction and opt-out requests reach this data; exceptions | [Automated search via EHR connector; HIPAA retention exception] |
| D6. Known issues | Gaps, open remediation items, owners, dates | [None] |
| D7. Attestation | Owner's yearly confirmation date | [2026-10-14] |
Sample: Privacy Impact Assessment Template
Sample only. Adapt it to your organization's risks, systems and laws, have qualified people review it, and do not treat it as a substitute for a risk assessment.
Follows: E-Government Act of 2002 Section 208 and OMB M-03-22 (Sept. 26, 2003); GDPR Article 35(7) and the WP248 rev.01 guidelines (as last revised Oct. 4, 2017); the content elements of the CalPrivacy risk assessment regulations (effective Jan. 1, 2026) as summarized in this lesson; Minnesota Statutes 325M.18; NIST Privacy Framework 1.0; Idaho Technology Authority guideline G502 for Idaho agencies. Where a law requires specific content, its text controls.
1. Project and decision
Project name: [ ]. Assessment ID: [ ]. Business owner: [name, title]. Privacy reviewer: [name]. Date started: [ ]. Proposed launch: [ ]. Type: [new system / new vendor / new use of existing data / change]. Screening triggers met: [sensitive data / sale or sharing / targeted advertising / automated decisions or profiling / AI training / large scale / monitoring / minors / cross-border / other].
2. Description of the processing
Purpose, stated specifically: [ ]. Data elements, with the reason each is needed and the minimum necessary: [table]. Categories and approximate number of people: [ ]. Sources: [ ]. How data is collected, used, disclosed and retained: [ ]. Systems and vendors, with their roles: [ ]. Recipients outside the organization: [ ]. Retention period and deletion method: [ ]. For automated decisions or AI: what the system does, its logic in plain terms, its known limits, and how humans use its output: [ ].
3. Necessity and proportionality
Why this purpose cannot be met with less data, less sensitive data, de-identified data or a shorter retention period: [ ]. Whether the use is compatible with what people were told when the data was collected: [ ].
4. Benefits
Benefits to the organization, to the people whose data is used, and to others: [ ].
5. Risks to people
For each risk, describe its source, likelihood and severity: unexpected use or disclosure; discrimination or unfair treatment; financial, physical or reputational harm; loss of control (inability to exercise rights); chilling effects from monitoring; security breach; harm to minors or other vulnerable people; inaccurate data leading to wrong decisions. [Risk register table with rating before safeguards.]
6. Safeguards
For each risk, the measures that reduce it (data fields dropped, retention shortened, access restricted, encryption, contract terms, consent or opt-out, notice changes, human review of automated outputs, testing for bias, tag restrictions), the owner and the date: [table with rating after safeguards].
7. Legal requirements checked
Laws and contracts reviewed: [list]. Notices to update: [ ]. Contracts to sign or amend: [ ]. Regulator consultation needed (GDPR Article 36 or other): [yes or no, why]. Legal advice received separately: [yes or no; do not summarize privileged advice here].
8. Contributors
Names and roles of everyone who provided information for this assessment: [ ].
9. Decision and approval
Decision: [proceed / proceed with conditions / do not proceed]. Conditions and due dates: [ ]. Residual risk accepted: [describe]. Approved by: [name, position, date; at least one approver with authority to decide whether the processing will start]. Privacy office concurrence: [name, date]. Next review: [no later than three years, or within 45 days of a material change, or sooner if required].
10. Record keeping
This assessment and all prior versions are kept for [the longer of the life of the processing or five years after completion, or a longer period if required]. Federal agencies: publish the PIA as required by Section 208 unless an exception applies.
Pictures, Numbers and Infographics
| Number or date | What it is | Source |
|---|---|---|
| Jan. 16, 2020 | NIST Privacy Framework Version 1.0 published; still the current final version on Nov. 15, 2026 | NIST2015 |
| April 14, 2025 | Privacy Framework 1.1 initial public draft released; comments to June 13, 2025 | NIST16 |
| Sept. 26, 2003 | OMB M-03-22, guidance for federal PIAs under E-Government Act Section 208 | OMB2 |
| 10 or more people | New federal collection of identifiable information that requires a PIA | OMB2 |
| 250 employees | GDPR Article 30(5) size threshold for the limited records exemption | GDPR1 |
| Jan. 1, 2026 | California risk assessment, cybersecurity audit and ADMT regulations take effect | CalPrivacy7 |
| Dec. 31, 2027 | Deadline to complete California risk assessments for processing already under way | Morgan Lewis summary19 |
| April 1, 2028 | First California risk assessment attestations and summaries due; first cybersecurity audit certifications due for businesses over $100 million revenue | CalPrivacy7 |
| April 1, 2029; April 1, 2030 | Audit certifications due for $50 million to $100 million and under $50 million | CalPrivacy7 |
| Jan. 1, 2027 | California ADMT requirements begin | CalPrivacy7 |
| 45 days | Deadline to update a California risk assessment after a material change | Morgan Lewis summary19 |
| 10,000,000 consumers | California threshold for publishing request metrics by July 1 each year | 11 CCR 710229 |
| 23 or 24 | State consumer privacy laws (IAPP, June 29, 2026, without Florida; MultiState, Oct. 28, 2026, with Florida) | IAPP; MultiState2726 |
| 2 | States with broad consumer privacy laws that do not require data protection assessments: Iowa and Utah | Sheppard Mullin chart6 |
| April 1, 2026 | Maryland's minimization rules apply to processing from this date (law effective Oct. 1, 2025) | Manatt9 |
| $10,000; $25,000 | Maryland civil penalty per violation; per repeat violation | Bloomberg Law commentary30 |
| $12.75 million | California AG settlement with GM, May 8, 2026, the largest CCPA penalty to date | Hunton; California AG831 |
| $7.8 million | BetterHelp payment for consumer refunds, FTC final order July 14, 2023 | FTC11 |
| $1.5 million | GoodRx civil penalty, Feb. 1, 2023 | FTC10 |
| June 20, 2024 | Court vacates part of HHS tracking guidance; HHS drops appeal Aug. 29, 2024 | Proskauer; HHS3212 |
| $632,500; $375,703; $1.10 million | CalPrivacy fines: Honda (March 12, 2025), Ford (March 2026), PlayOn Sports (March 2026) | CalPrivacy141334 |
| April 22, 2026 | Compliance date for most of the amended COPPA Rule | 90 FR 1691817 |
| $1.375 billion | Texas settlement with Google, May 2025 | Texas Lawbook37 |
The program wheel and data map earlier in this lesson, and the assessment trigger flow in the assessments section, are the other infographics for this lesson.
Common mistakes
Privacy promises that do not match the code
GoodRx and BetterHelp both told users their health information would be kept private or used only for limited purposes, while tracking tools on their sites and apps sent it to advertising platforms.1011 Under Section 5 of the FTC Act, a privacy statement that is not true is a deceptive practice.22 The fix is procedural: no privacy notice is approved without checking it against the inventory and a live scan of tags, and no tag is added without checking it against the notice.
Making rights hard to use
Honda asked for more information than needed to process requests and required verification for opt-outs; Ford required email verification before opt-outs; PlayOn Sports sent people to industry opt-out pages instead of providing its own.141334 Each choice probably looked like caution or convenience to the team that made it. Each was a violation. A program should test its own request flows the way a regulator would, with a stopwatch and a count of steps.
Ignoring browser opt-out signals
Sephora's 2022 settlement, and later the Ford and PlayOn decisions, found failures to honor Global Privacy Control signals.281334 Honoring the signal is a technical setting in most consent platforms; the failure is usually that nobody owns checking it.
Using data for a new purpose without asking
California's GM settlement turned on data collected for connected vehicle services being sold for other purposes, including to data brokers that serve insurers.8 A purpose field in the inventory, checked whenever a business team proposes a new use or a new recipient, is the control that catches this before it happens.
Tracking on sensitive pages
Hospitals, telehealth companies and health information sites put advertising tags on pages about conditions, appointment scheduling and patient portals. Even after the court cut back HHS's guidance in 2024, the parts of it on authenticated portals remain, and the FTC and California cases show that health-adjacent data outside HIPAA is treated as sensitive.3233
Keeping everything forever
Breach notices from large organizations regularly include people who stopped being customers or employees many years earlier. Old data raises breach costs, rights request workloads and discovery costs, and it now conflicts with minimization rules in Maryland and California.921 The BetterHelp order's retention schedule requirement shows what regulators expect.11
Assessments written after the decision
An assessment completed after launch to fill a file protects nothing, and California's rule that the record name a person with authority to decide whether the processing will start, along with the contributors and dates, makes a backdated or rubber-stamp assessment visible.19 Programs should measure how many assessments were finished before launch.
Questions people ask
Do we need a chief privacy officer?
No U.S. law outside specific sectors requires that title, but several require a named person: a HIPAA privacy official, a GDPR data protection officer in some cases, a federal agency's Senior Agency Official for Privacy, a COPPA security program coordinator, and in Minnesota a contact for the person with primary responsibility for privacy compliance, where applicable.14175 A large organization handling consumer data in many states needs someone whose full-time job is the program; what the title is matters less than the authority.
Is the NIST Privacy Framework required?
No. It is voluntary. Its value is as a structure for a program and a common vocabulary with the Cybersecurity Framework. Version 1.0 is the current final version; Version 1.1 was still a draft on Nov. 15, 2026.15
Can one assessment satisfy several laws?
Usually, if it contains every element each law requires. Most state laws accept an assessment done for another law if it is reasonably comparable in scope and effect, and a template built on the strictest content (California's, with its approver and contributor fields, and GDPR Article 35(7)) can satisfy the others with a short supplement.191 Check the specific law before relying on reuse.
Will our assessments become public?
For private companies, state laws generally let the attorney general obtain assessments but keep them confidential and exempt from public records laws, and California requires summaries, not full assessments, to be submitted. For government agencies, the answer is different: federal PIAs are published unless an exception applies, and state and local agency assessments may be public records, subject to exemptions, under laws such as the Idaho Public Records Act.241 Write every assessment as if it will be read by the people it describes.
How big should the privacy team be?
There is no regulatory standard. The workload is driven by the number of systems and vendors, request volume, the number of states and countries, the pace of new products, and whether the organization is under an order. A practical test is whether assessments are completed before launch, requests are answered on time, and the inventory is attested each year; if those slip, the team is too small or the process is wrong. Budget arguments are covered in lesson 2.
Does Idaho have a consumer privacy law?
Not a broad one as of the 2026 session. Idaho has a breach notification law (Idaho Code 28-51-104 to 28-51-107) covered in lesson 108-11, privacy rules for state agencies through the Idaho Technology Authority, and sector laws such as those on student data. Idaho organizations with customers in other states must follow those states' laws, and Idaho's large employers, hospitals and insurers face HIPAA, Gramm-Leach-Bliley and the FTC Act regardless.18
How does the privacy program connect to AI governance?
Closely. California's risk assessment triggers include using automated decisionmaking for significant decisions and training such technology on personal data, and the ADMT rules begin Jan. 1, 2027.197 Most organizations run one intake for AI and privacy review and keep the AI inventory linked to the data inventory. Lesson 13 covers the AI side.
Hard questions
Does strict minimization hold back useful products?
The case for strict limits: data that is not collected cannot be breached, sold or misused, and the GM and BetterHelp cases show that data collected for one purpose drifts to others when it is available. Maryland's "strictly necessary" standard for sensitive data puts the burden where the risk is, on the organization that wants the data.98 The case for flexibility: many valuable uses (fraud detection, safety research, product improvement, medical research) are discovered after collection, and a rule tied to the service the consumer requested may block them or push companies toward consent screens that people click through without reading. What the evidence shows: there is not yet enforcement history under Maryland's law to show how strictly it will be read, and the California GM case involved sales to data brokers, not internal research. The practical middle path most programs take is to document compatible purposes in advance, de-identify where possible, and run an assessment before any new use.
Do privacy assessments change decisions, or just produce paper?
The case that they work: an assessment forces a team to state its purpose and list its data before launch, and the most common outcome is a smaller data set or shorter retention, which is a real reduction in risk. The approver requirement in California's rules puts a named executive on the record.19 The case that they are paperwork: assessments are often done late, by the team that wants to proceed, and almost never end in "do not proceed"; regulators rarely read them unless something goes wrong. What the evidence shows: there is little public data on outcomes. Programs can produce their own evidence by tracking how many assessments resulted in changes, how many were completed before launch, and how many conditions were actually carried out, and boards should ask for those numbers.
Should a federal law replace the state patchwork?
The case for a federal law: large organizations now track 20 or more state laws with different definitions, thresholds and deadlines, which raises costs without clearly improving protection, and people's rights should not depend on their address. The case for keeping state laws: states have moved faster than Congress and produced stronger rules (Maryland's minimization, California's agency and audits); a federal law that preempts them could lower protection for many people. What the evidence shows: Congress has considered bills for years without passing one. As lesson 108-12 reports, the SECURE Data Act, introduced April 21, 2026 with preemption and no private right of action, had a subcommittee hearing on June 3, 2026 and no recorded votes as of mid-November 2026.42 Programs should plan on state laws remaining the operative rules for the foreseeable future.
Is a visit to a public health page health information?
HHS's position in 2022 and 2024: an IP address collected when someone visits a page about a specific condition or provider can reveal that the person is, or may become, a patient, and sending it to advertisers discloses protected health information. The court's view: in June 2024 the Northern District of Texas held that combining an IP address with a visit to an unauthenticated public page does not meet HIPAA's definition of individually identifiable health information, and vacated that part of the guidance.32 What it means: the HIPAA question for public pages is settled for now by the court's ruling, which HHS did not appeal, but the FTC, state privacy laws (most of which treat health inferences as sensitive data) and wiretap class actions still reach the same conduct. Hospitals that removed advertising pixels from health pages in 2023 and 2024 have, for the most part, not put them back.
For reporters and advocates
What public records exist.
- Privacy policies and request metrics. Companies that handle data on 10 million or more Californians must publish rights request metrics by July 1 each year, in or linked from their privacy policy.29 Compare a company's numbers year to year and against competitors: denial rates and response times are telling. Archived copies of privacy policies show when promises changed.
- Federal PIAs and system of records notices. Agencies publish privacy impact assessments on their websites under E-Government Act Section 208 and system of records notices in the Federal Register.2 A new agency system without a PIA is a story.
- Enforcement records. The FTC posts complaints, orders and later compliance reports; the California attorney general and CalPrivacy post settlements and board decisions; the Texas attorney general posts press releases and petitions.10311437 Court dockets for state and FTC cases are on PACER (many mirrored free on CourtListener) and state court systems.
- Breach records. The HHS breach portal lists health breaches affecting 500 or more people, and many state attorneys general post breach notices; Idaho's attorney general lists notices received since 2021.4344 A breach that included people who left the organization years ago points to a retention failure.
- SEC filings. Public companies' 10-K risk factors increasingly describe privacy laws and enforcement; Item 1C covers cyber risk management (lesson 108-04).
- Your own browser. Anyone can see which third-party domains a public website contacts using the browser's built-in developer tools or a privacy-focused browser extension, without breaking any rule. Hospital, school district and government sites are fair subjects.
- Idaho public records. Under the Idaho Public Records Act (Idaho Code Title 74, Chapter 1), a person can request a state or local agency's privacy policies, privacy impact assessments under ITA guideline G502, vendor contracts with data terms, and records of the agency's website analytics and advertising contracts, subject to the act's exemptions.4118
What to ask an organization.
- Who is your chief privacy officer, to whom do they report, and when did they last brief the board?
- Do you have a written data inventory, and when was it last attested by its owners?
- How many privacy assessments did you complete last year, how many were finished before launch, and how many changed the project?
- How many rights requests did you receive, what share were answered on time, and do you honor Global Privacy Control?
- Which third-party tracking tools run on your website and app, and are any on pages about health, finances or children?
- What is your retention period for former customers' and employees' data, and does deletion actually run?
- Have you sold or shared data with data brokers or insurers in the last three years?
- For a government agency: where is the PIA for this system, and is it published?
The Idaho angle. Idaho has no broad consumer privacy law, so Idaho residents have fewer statutory rights than residents of neighboring Oregon, Montana and Utah, while Idaho companies serving those states must comply anyway. Legislators and county commissioners can ask whether state and local agencies follow ITA's privacy policy and PIA guideline, whether school districts and hospital districts review tracking on their websites, and how the state would respond to a federal law that preempts or does not preempt state rules.
Where to go next
- NIST Privacy Framework: NIST's free, voluntary privacy risk framework with Version 1.0, the 1.1 draft, crosswalks and a quick-start guide. For privacy officers building or measuring a program. Free. Start with a Current Profile of the Identify-P inventory and mapping outcomes.
- CalPrivacy regulations and enforcement: California's privacy agency: the approved text of the risk assessment, cybersecurity audit and ADMT rules, plus enforcement decisions. For any business meeting the CCPA thresholds. Free. Read the decisions; they show exactly what the agency tests.
- OMB M-03-22 (federal PIA guidance): The federal government's guidance on privacy impact assessments under E-Government Act Section 208. For federal agencies, contractors and anyone designing a PIA template. Free. Use its content list as a checklist.
- Idaho Technology Authority policies: Idaho's privacy policy (P4150), PIA guideline (G502) and classification policies for state agencies. For Idaho public officials and vendors to the state. Free. Check the current version dates before relying on them.
- FTC privacy and security business guidance: FTC guidance on COPPA, health apps, the Health Breach Notification Rule and data security. For compliance officers, marketers and product teams. Free. Read with the GoodRx and BetterHelp orders.
- HHS OCR breach portal: Public list of health data breaches affecting 500 or more people. For reporters, advocates and health compliance teams. Free. Search by entity or state.
Still being checked (6 items)
These points could not be fully confirmed against a primary source when this lesson was written and fact checked in November 2026. The text above states each one as uncertain.
- The count of states requiring data protection assessments relies on a Sheppard Mullin chart that does not show an update date; the assessment provisions of the four 2026 laws (Alabama, Louisiana, Oklahoma, Vermont) were not checked.
- The exact dates of CalPrivacy's Ford and PlayOn Sports decisions were confirmed only to March 2026.
- Whether the 2025 CCPA rulemaking amended 11 CCR 7102 was not confirmed; the Legal Information Institute text shows amendments through March 29, 2023.
- The 24-month request record rule (11 CCR 7101) and the two-method rule for access and deletion requests were not re-read on Nov. 15, 2026; the opt-out rules (two or more methods, opt-out preference signals, and no verification for opt-outs) were confirmed in 11 CCR 7026(a) and (d).
- The status of the Texas suit against Allstate and Arity after 2025 was not confirmed.
- The Bloomberg Law commentary on Maryland penalties is undated in the source read.
Last verified November 15, 2026. General education, not legal, compliance, audit or other professional advice; rule versions and dates are stated in the text, and the sources used are listed in the Notes and Sources below. Check the regulator's current text before relying on any point.
Notes
- Regulation (EU) 2016/679, General Data Protection Regulation, Official Journal L 119, May 4, 2016, Articles 5, 12, 25, 30, 35, 36, 37 and 38, EUR-Lex (read Nov. 15, 2026). Read it online. ↩
- Office of Management and Budget, Memorandum M-03-22, OMB Guidance for Implementing the Privacy Provisions of the E-Government Act of 2002, Sept. 26, 2003, implementing Section 208 of the E-Government Act of 2002 (Pub. L. 107-347) (read Nov. 15, 2026). Read it online. ↩
- California Privacy Protection Agency (CalPrivacy), CCPA Updates, Cybersecurity Audits, Risk Assessments, Automated Decisionmaking Technology and Insurance Regulations page: adopted July 24, 2025, approved by the Office of Administrative Law and filed Sept. 22, 2025, effective Jan. 1, 2026, with the approved regulation text (11 CCR 7000 and following) (read Nov. 15, 2026). Read it online. ↩
- Office of Management and Budget, Circular A-130, Managing Information as a Strategic Resource, July 2016: sections 5.a.1 (PII inventory) and 5.f.1 (Senior Agency Official for Privacy; agency privacy program), and the definitions of privacy program plan and privacy continuous monitoring; Executive Order 13719 (Feb. 9, 2016) (read Nov. 15, 2026). Read it online. ↩
- Minnesota Statutes 325M.18 (2026), Data Privacy Policies; Data Privacy and Protection Assessments, Minnesota Consumer Data Privacy Act, Office of the Revisor of Statutes (read Nov. 15, 2026). Read it online. ↩
- Sheppard Mullin, U.S. State Comprehensive Privacy Laws comparison chart (undated web chart): data protection assessments must be documented in all covered states except Iowa and Utah (read Nov. 15, 2026). Read it online. ↩
- California Privacy Protection Agency, California Finalizes Regulations to Strengthen Consumers' Privacy, announcement, Sept. 23, 2025, with compliance dates for risk assessments (from Jan. 1, 2026; submissions by April 1, 2028), cybersecurity audit certifications (April 1, 2028, 2029 and 2030 by revenue) and ADMT (from Jan. 1, 2027) (read Nov. 15, 2026). Read it online. ↩
- Hunton Andrews Kurth, California AG Announces Record $12.75M Settlement with GM over CCPA Data Minimization and Purpose Limitation Violations, May 14, 2026 (settlement announced May 8, 2026, subject to court approval) (read Nov. 15, 2026). Read it online. ↩
- Manatt, Phelps and Phillips, Now in Effect: Maryland Law Raises Bar on Sensitive Data, Data Minimization and Children's Privacy, Oct. 2, 2025 (Maryland Online Data Privacy Act effective Oct. 1, 2025; applies to processing on or after April 1, 2026; thresholds; minimization standards; sale and under-18 bans) (read Nov. 15, 2026). Read it online. ↩
- Federal Trade Commission, FTC Enforcement Action to Bar GoodRx from Sharing Consumers' Sensitive Health Info for Advertising, press release, Feb. 1, 2023 ($1.5 million civil penalty) (read Nov. 15, 2026). Read it online. ↩
- Federal Trade Commission, FTC Gives Final Approval to Order Banning BetterHelp from Sharing Sensitive Health Data for Advertising, Requiring It to Pay $7.8 Million, press release, July 14, 2023 (read Nov. 15, 2026). Read it online. ↩
- U.S. Department of Health and Human Services, Office for Civil Rights, Use of Online Tracking Technologies by HIPAA Covered Entities and Business Associates, with notice of the June 20, 2024 order in American Hospital Association v. Becerra (content last reviewed June 26, 2024) (read Nov. 15, 2026). Read it online. ↩
- CalPrivacy, Ford to Change Practices, Pay Fine for Adding Unnecessary Friction to Opt-Out Process, March 2026 ($375,703; audit of tracking technologies and Global Privacy Control compliance required) (read Nov. 15, 2026). Read it online. ↩
- California Privacy Protection Agency, announcement of the American Honda Motor Co. decision, March 12, 2025 ($632,500) (read Nov. 15, 2026). Read it online. ↩
- National Institute of Standards and Technology, NIST Privacy Framework page: Version 1.0 listed as the framework document; Version 1.1 labeled an initial public draft (page last modified June 24, 2026) (read Nov. 15, 2026). Read it online. ↩
- NIST, NIST Privacy Framework 1.1, initial public draft, NIST CSWP 40 ipd, April 14, 2025, comments due June 13, 2025; changes summarized in Hunton Andrews Kurth, NIST Releases Updated Privacy Framework, April 2025 (read Nov. 15, 2026). Read it online. ↩
- Federal Trade Commission, Children's Online Privacy Protection Rule, final rule, 90 FR 16918 (April 22, 2025), amending 16 CFR Part 312; effective June 23, 2025, compliance by April 22, 2026 for most provisions (read Nov. 15, 2026). Read it online. ↩
- Idaho Technology Authority, Policies, Standards and Guidelines page, listing P4150 Privacy Policy, G502 Privacy Impact Assessment, P4130 Information Systems Classification Policy and G505 Data Classification and Labeling (read Nov. 15, 2026). Read it online. ↩
- Morgan Lewis, CCPA Risk Assessment Requirements and Best Practices, July 2026: the six significant-risk triggers, nine content elements, Dec. 31, 2027 deadline for existing processing, three-year review, 45-day update, retention and April 1, 2028 submission (read Nov. 15, 2026). Read it online. ↩
- NIST, The NIST Privacy Framework: A Tool for Improving Privacy Through Enterprise Risk Management, Version 1.0, NIST CSWP 01162020, Jan. 16, 2020 (read Nov. 15, 2026). Read it online. ↩
- California Civil Code 1798.100 to 1798.199.100, California Consumer Privacy Act of 2018 as amended by the California Privacy Rights Act of 2020, including 1798.100 (notice, retention disclosure and the reasonably necessary and proportionate standard in subdivision (c)), 1798.130 (methods and 45-day response period) and 1798.145 (former employee data exemption, expired Jan. 1, 2023), California Legislative Information (read Nov. 15, 2026). Read it online. ↩
- 15 U.S.C. 45, Federal Trade Commission Act Section 5, unfair or deceptive acts or practices, Legal Information Institute text (read Nov. 15, 2026). Read it online. ↩
- U.S. Department of Justice, National Security Division, Preventing Access to U.S. Sensitive Personal Data and Government-Related Data by Countries of Concern or Covered Persons, final rule, Federal Register document 2024-31486, Jan. 8, 2025, codified at 28 CFR Part 202, effective April 8, 2025 (read Nov. 15, 2026). Read it online. ↩
- Lewis Silkin, Digital rulebook: simplification of EU data protection and cyber rules, Nov. 9, 2026, on the Council of the EU position agreed Oct. 21, 2026 (read Nov. 15, 2026). Read it online. ↩
- Article 29 Data Protection Working Party, Guidelines on Data Protection Impact Assessment (DPIA) and determining whether processing is "likely to result in a high risk", WP248 rev.01, adopted April 4, 2017, as last revised and adopted Oct. 4, 2017 (posted by the European Commission Oct. 13, 2017), endorsed by the European Data Protection Board (read Nov. 15, 2026). Read it online. ↩
- MultiState, Twenty-four states now have comprehensive consumer privacy laws (2026 State Privacy Law Trends), Oct. 28, 2026, on the Alabama, Louisiana, Oklahoma and Vermont enactments and 2026 amendments including Delaware's profiling assessments (read Nov. 15, 2026). Read it online. ↩
- International Association of Privacy Professionals, US State Privacy Legislation Tracker 2026, chart of signed laws, last updated June 29, 2026 (read Nov. 15, 2026). Read it online. ↩
- California Attorney General, Attorney General Bonta Announces Settlement with Sephora as Part of Ongoing Enforcement of California Consumer Privacy Act, press release, Aug. 24, 2022 ($1.2 million) (read Nov. 15, 2026). Read it online. ↩
- Cal. Code Regs. tit. 11, section 7102, Requirements for Businesses Collecting Large Amounts of Personal Information, renumbered from 999.317(g) and (h) in 2022 and amended March 29, 2023, Legal Information Institute text. This lesson did not find a 2025 change to the section's 10 million threshold (read Nov. 15, 2026). Read it online. ↩
- Bloomberg Law, Maryland Privacy Crackdown Raises Bar for Disclosure Compliance, commentary by Labaton Sucharow attorneys, 2025: civil penalties up to $10,000 per violation and $25,000 for repeat violations; enforcement from April 1, 2026 (read Nov. 15, 2026). Read it online. ↩
- California Attorney General, Privacy Enforcement Actions page, listing General Motors ($12,750,000, May 8, 2026), Disney ($2,750,000, Feb. 11, 2026), Jam City ($1,400,000, Nov. 21, 2025), Illuminate Education ($3.25 million, Nov. 6, 2025), Sling TV and Dish Media Sales ($530,000, Oct. 30, 2025) and Healthline ($1.55 million, July 1, 2025) (read Nov. 15, 2026). Read it online. ↩
- Proskauer Rose, Shortly After Its Online Tracking Technologies Bulletin Is Declared Unlawful, HHS-OCR Stands Down, Withdraws Appeal, 2024: Am. Hosp. Ass'n v. Becerra, No. 4:23-cv-1110 (N.D. Tex. June 20, 2024); appeal withdrawn Aug. 29, 2024 (read Nov. 15, 2026). Read it online. ↩
- California Attorney General, Attorney General Bonta Announces Largest CCPA Settlement to Date, Secures $1.55 Million from Healthline.com, press release, July 1, 2025 (read Nov. 15, 2026). Read it online. ↩
- CalPrivacy, Youth Sports Media Company to Pay $1.1 Million Fine, Change Practices over Privacy Violations (PlayOn Sports), March 2026 (read Nov. 15, 2026). Read it online. ↩
- Federal Trade Commission, Fortnite Video Game Maker Epic Games to Pay More Than Half a Billion Dollars over FTC Allegations of Privacy Violations and Unwanted Charges, press release, Dec. 19, 2022 ($275 million COPPA penalty) (read Nov. 15, 2026). Read it online. ↩
- National Law Review, Texas Attorney General Reaches Largest Ever Biometrics Settlement with Meta, Aug. 28, 2024 (settlement order July 30, 2024, $1.4 billion, under the Capture or Use of Biometric Identifier Act) (read Nov. 15, 2026). Read it online. ↩
- Texas Lawbook, Texas Emerges as a Leading Force in State Privacy Law Enforcement, Nov. 20, 2025: Meta ($1.4 billion, July 2024), Allstate and Arity suit (January 2025), DeepSeek investigation (February 2025), Google ($1.375 billion, May 2025) (read Nov. 15, 2026). Read it online. ↩
- Kelley Drye, AG Watch: Texas Is Entering New Privacy Enforcement Era, March 13, 2025 (Law360 article on the Allstate and Arity suit, the first under the Texas Data Privacy and Security Act) (read Nov. 15, 2026). Read it online. ↩
- California Privacy Protection Agency, Nation's Largest Rural Lifestyle Retailer to Pay $1.35M Over CCPA Violations (Tractor Supply Company), Sept. 30, 2025 (read Nov. 15, 2026). Read it online. ↩
- California Privacy Protection Agency (CalPrivacy), CalPrivacy Brings First Action Against a Data Broker Under Both the CCPA and Delete Act, August 2026 (LocateSmarter LLC, $116,490) (read Nov. 15, 2026). Read it online. ↩
- Idaho Code Title 74, Chapter 1, Public Records Act, Idaho Legislature (read Nov. 15, 2026). Read it online. ↩
- LegiScan, H.R. 8413 (119th Congress), SECURE Data Act: introduced April 21, 2026, referred to Energy and Commerce and Judiciary, no recorded votes; House Energy and Commerce subcommittee legislative hearing June 3, 2026, as reported in Module 108 lesson 12 (checked Nov. 13, 2026) (read Nov. 15, 2026). Read it online. ↩
- U.S. Department of Health and Human Services, Office for Civil Rights, breach portal (breaches of unsecured protected health information affecting 500 or more individuals) (read Nov. 15, 2026). Read it online. ↩
- Idaho Attorney General, Consumer Protection Division, Security Breaches page, with notices received since Jan. 1, 2021 (read Nov. 15, 2026). Read it online. ↩
Check yourself
1. What is the current final version of the NIST Privacy Framework as of Nov. 15, 2026?
NIST's Privacy Framework page, last modified June 24, 2026, lists Version 1.0 (NIST CSWP 01162020, Jan. 16, 2020) as current; Version 1.1 (CSWP 40) is an initial public draft released April 14, 2025.
2. Under E-Government Act Section 208 as implemented by OMB M-03-22, which of these requires a federal agency PIA?
OMB M-03-22 (Sept. 26, 2003) requires a PIA before developing or procuring IT that collects identifiable information or initiating a new collection of identifiable information from 10 or more persons (other than federal employees).
3. Which of these is NOT one of the items GDPR Article 30(1) requires in a controller's record of processing?
Article 30(1) lists the controller's details, purposes, categories of data subjects and data, recipients, third-country transfers, erasure time limits where possible and a general description of security measures; revenue is not among them.
4. Under California's risk assessment regulations effective Jan. 1, 2026, by when must a business submit its first attestation and summary to the agency?
CalPrivacy's Sept. 23, 2025 announcement states that businesses must begin compliance Jan. 1, 2026 and submit an attestation and summary of risk assessment information by April 1, 2028.
5. Which two states' broad consumer privacy laws, according to the Sheppard Mullin comparison chart, do not require documented data protection assessments?
The Sheppard Mullin chart shows assessments must be documented in all covered states except Iowa and Utah.
6. What did California's $12.75 million settlement with General Motors, announced May 8, 2026, primarily rest on?
Hunton Andrews Kurth (May 14, 2026) reports the settlement was the first CCPA case focused on data minimization and purpose limitation, involving sales of driving data to LexisNexis Risk Solutions and Verisk.
7. What did the court decide in American Hospital Association v. Becerra on June 20, 2024?
The Northern District of Texas declared unlawful and vacated the 'Proscribed Combination' portion of the HHS-OCR bulletin; HHS withdrew its appeal Aug. 29, 2024 (Proskauer; HHS notice).
8. Why did CalPrivacy fine Ford $375,703 in March 2026?
CalPrivacy's March 2026 announcement says Ford required consumers to verify their email address before opting out, creating unnecessary friction, and ordered an audit of tracking technologies and GPC compliance.
9. Under 11 CCR 7102, which businesses must publish annual rights request metrics by July 1?
11 CCR 7102(a) applies to businesses that buy, receive, sell, share or otherwise make available personal information of 10,000,000 or more consumers in a calendar year.
10. Maryland's Online Data Privacy Act limits processing of sensitive data to what is:
As summarized by Manatt (Oct. 2, 2025), Maryland limits sensitive data processing to what is 'strictly necessary' for a requested product or service, and consent does not override the limit; the law applies to processing from April 1, 2026.
Terms in this lesson
- Chief privacy officer (CPO)
- The executive who owns an organization's privacy program: policies, inventory, assessments, rights requests, training and reporting.
- Senior Agency Official for Privacy (SAOP)
- The official each federal agency must designate under Executive Order 13719 and OMB Circular A-130 with agency-wide responsibility for privacy.
- Data protection officer (DPO)
- A role the GDPR requires for public authorities and for organizations whose core activities involve large-scale monitoring or sensitive data; protected from dismissal for doing the job and reporting to top management.
- Privacy champion
- A person in a business unit, usually part-time, trained to spot privacy triggers, keep the unit's inventory current and route requests to the central privacy office.
- Data inventory
- A written list of the personal data an organization holds, by system or process, with owners, purposes, recipients and retention.
- Record of processing activities (ROPA)
- The GDPR Article 30 record of each processing activity: purposes, categories of people and data, recipients, transfers, erasure time limits and security measures.
- Data map
- A diagram showing how personal data moves from people through systems to recipients and finally to deletion.
- Privacy impact assessment (PIA)
- A written review, done before a project begins, of the personal data involved, the risks to people and the measures to reduce them; required of federal agencies by E-Government Act Section 208.
- Data protection impact assessment (DPIA)
- The GDPR Article 35 assessment required before processing likely to result in a high risk to people's rights and freedoms.
- Automated decisionmaking technology (ADMT)
- Under California's regulations, technology that processes personal information to replace or substantially replace human decisionmaking; subject to notice, opt-out and access rules from Jan. 1, 2027 when used for significant decisions.
- Data minimization
- Collecting, using and keeping only the personal data reasonably necessary for a disclosed purpose; a legal requirement in California, Maryland and the GDPR.
- Global Privacy Control (GPC)
- A browser setting that sends an opt-out preference signal to websites; California and several other states require businesses to treat it as a valid opt-out of sale or sharing.
- Tracking pixel
- A piece of third-party code on a website or in an email that reports user activity and identifiers to an analytics or advertising company.
- Retention schedule
- A list of records categories with the reason to keep each, the retention period, the event that starts the clock and the disposal method.
Sources
- NIST Privacy Framework Version 1.0 (CSWP 01162020) (2020-01-16) checked 2026-11
- NIST Privacy Framework 1.1 initial public draft (CSWP 40 ipd) (2025-04-14) checked 2026-11
- General Data Protection Regulation, Regulation (EU) 2016/679 (2016-05-04) checked 2026-11
- OMB M-03-22, Guidance for Implementing the Privacy Provisions of the E-Government Act of 2002 (2003-09-26) checked 2026-11
- OMB Circular A-130, Managing Information as a Strategic Resource (2016-07-28) checked 2026-11
- CalPrivacy, regulations on cybersecurity audits, risk assessments and ADMT (2025-09-22) checked 2026-11
- California Privacy Protection Agency announcement of final regulations (2025-09-23) checked 2026-11
- Cal. Code Regs. tit. 11, section 7102 (2023-03-29) checked 2026-11
- Minnesota Statutes 325M.18 (2026-01-01) checked 2026-11
- FTC COPPA Rule final amendments, 90 FR 16918 (2025-04-22) checked 2026-11
- FTC, BetterHelp final order press release (2023-07-14) checked 2026-11
- HHS OCR, Use of Online Tracking Technologies by HIPAA Covered Entities and Business Associates (2024-06-26) checked 2026-11
- California Attorney General, Privacy Enforcement Actions (2026-05-08) checked 2026-11
- Idaho Technology Authority, Policies, Standards and Guidelines (2026-11-15) checked 2026-11