Topics
14 compliance topics, from deadlines to insurance
| Topic | What it covers |
|---|---|
| Regulations and Standards | Which AI compliance duties take effect when. |
| EU AI Act Conformity | How high-risk AI and general-purpose models show they meet the AI Act. |
| Notified Bodies | The third parties that will certify high-risk AI in the EU, and their counterparts elsewhere. |
| UL 4600 | The safety case standard for autonomous vehicles and how regulators use it. |
| ISO/IEC 42001 | The certifiable management system standard for AI. |
| NIST AI RMF | How agencies and companies apply the NIST AI Risk Management Framework. |
| Sector Compliance Gaps | Where sector rules for AI agents are missing or uneven. |
| Third-Party Audits | Independent testing and assurance of AI systems. |
| AI Audits | Internal audit of AI systems and the logs that make audit possible. |
| Human-in-the-Loop Oversight | Human review as a legal requirement, and whether it works. |
| Conformance | Testing and certification schemes that show an AI product meets a standard. |
| Auditable Documentation | The records AI rules require developers and deployers to keep. |
| Penalties and Fines | What AI-related enforcement has cost companies so far. |
| Insurance and Underwriting | How insurers cover, price and exclude AI risk. |
Enforcement
28 enforcement actions: US regulators use consumer protection law, Europe uses privacy law, China uses takedowns
Amounts are as announced; the OpenAI fine is listed but excluded from totals because a court annulled it. Actions without a penalty (bans, orders, takedowns) are included.
| Date | Regulator | Country | Company | Penalty | What for | Status | Source |
|---|---|---|---|---|---|---|---|
| 2026-08-26 | FTC | United States | Cox Media Group, MindSift, 1010 Digital Works | $930,000 | "Active Listening" ad-targeting claims for a capability that did not exist | Final 20-year orders | Source |
| 2026-07-09 | Garante | Italy | Character Technologies (Character.AI) | €158,000 | Weak age verification and protection of minors | Final | Source |
| 2026-07-06 | CAC | China | Multiple AI products | None | 14,000+ AI products and 26,000+ accounts handled | Second phase announced | Source |
| 2026-01-27 | FTC | United States | Growth Cave | Not stated | False claims its AI software automates nearly all work | Proposed order | Source |
| 2026-01-26 | European Commission | EU | X (Grok) | None so far | Sexualized deepfakes made with Grok | Investigation open | Source |
| 2025-08-01 | FTC | United States | Air AI | Not stated | Deceptive claims about its AI agents | Pending | Source |
| 2025-06-20 | CAC | China | Multiple AI products | None | First phase handled 3,500+ AI products | Completed | Source |
| 2025-05-09 | Texas Attorney General | United States | $1.375B | Biometric, location and incognito data | Settled | Source | |
| 2025-04-28 | FTC | United States | Workado | None | Unsupported accuracy claim for an AI-content detector | Resolved | Source |
| 2025-04-23 | PIPC | South Korea | DeepSeek | None | Transfers of user data to China and the US | Corrective recommendations; app relisted | Source |
| 2025-04-10 | Garante | Italy | Luka Inc. (Replika) | €5M | No legal basis for processing; no age verification | Final | Source |
| 2025-01-30 | Garante | Italy | DeepSeek | None | Did not answer data-processing questions | Ban continues | Source |
| 2024-12-20 | Garante | Italy | OpenAI | €15M | Unlawful processing for training; no age checks | Annulled by Rome court Mar 18, 2026 | Source |
| 2024-12-03 | FTC | United States | IntelliVision | None | Unsupported claims of bias-free facial recognition | Order | |
| 2024-11-26 | FTC | United States | Evolv Technologies | None | Overstated AI weapons-detection claims | Order | |
| 2024-11-14 | DOJ | United States | Cruise | $500,000 | Misled the crash investigation | Deferred prosecution agreement | |
| 2024-09-30 | NHTSA | United States | Cruise | $1.5M | Incomplete crash reports | Consent order | |
| 2024-09-25 | FTC | United States | DoNotPay | $193,000 | Unsupported "robot lawyer" claims | Final order Feb 2025 | Source |
| 2024-09-18 | Texas Attorney General | United States | Pieces Technologies | None | Overstated accuracy of a hospital generative AI tool | Five-year assurance of compliance | |
| 2024-09-03 | Dutch DPA | Netherlands | Clearview AI | €30.5M | Illegal facial-image database | Contested | Source |
| 2024-07-30 | Texas Attorney General | United States | Meta | $1.4B | Facial recognition without consent | Settled | Source |
| 2024-03-20 | Autorité de la concurrence | France | €250M | Included training Bard on news content without notice | Settled | Source | |
| 2024-03-18 | SEC | United States | Delphia | $225,000 | False claims about its use of AI | Settled | |
| 2024-03-18 | SEC | United States | Global Predictions | $175,000 | False claims about its use of AI | Settled | |
| 2024-02-14 | BC Civil Resolution Tribunal | Canada | Air Canada | CAD 812 | Chatbot gave wrong bereavement-fare information | Ordered to pay | |
| 2023-12-19 | FTC | United States | Rite Aid | None | Reckless use of facial recognition in stores | Five-year facial recognition ban | Source |
| 2023-10-24 | California DMV | United States | Cruise | None | Driverless permits suspended after a pedestrian was dragged | Cruise ended robotaxi program Dec 2024 | |
| 2023-05-31 | FTC and DOJ | United States | Amazon (Alexa) | $25M | Kept children's voice recordings, partly to train algorithms | Settled | Source |
Evidence
Governance programs are common; agent governance lags agent use
| Date | Development | Country | What happened | Source |
|---|---|---|---|---|
| May 21, 2025 | IAPP AI Governance Profession Report | Global | 77% of organizations working on AI governance; talent shortage a top challenge | Source |
| Dec 8, 2025 | McKinsey State of AI 2025 | Global | 62% experimenting with agents, 23% scaling them; average of four risks mitigated | Source |
| Dec 31, 2025 | ISO Survey | Global | 271 accredited ISO/IEC 42001 certificates worldwide in the first year counted | Source |
| Jul 10, 2026 | CAC filing totals | China | 988 generative AI services filed and 598 apps registered | Source |
| Aug 29, 2026 | First EU AI Office information requests | EU | Requests to frontier model providers and 30+ firms on transparency and copyright | Source |
Registry
Nine registry incidents bear on AI compliance
| Incident | Date | Regulator or court action |
|---|---|---|
| Uber Test Vehicle Kills Pedestrian in Tempe | Mar 2018 | Arizona suspended Uber's testing on March 26, 2018. Yavapai County declined to charge Uber in March 2019. NTSB adopted its findings on November 19, 2019. The backup operator pleaded guilty to endangerment in July 2023 and received three years of supervised probation. |
| Amazon Echo Sends a Private Conversation to a Contact | May 2018 | None found. |
| Air Canada Chatbot Invents a Refund Rule | Nov 2022 | On February 14, 2024, the BC Civil Resolution Tribunal found negligent misrepresentation and ordered Air Canada to pay CA$812.02. |
| Cruise Robotaxi Drags a Pedestrian | Oct 2023 | California DMV suspended Cruise's driverless permits on October 24, 2023. NHTSA fined Cruise $1.5 million on September 30, 2024, for incomplete crash reports. On November 14, 2024, Cruise admitted filing a false report and agreed to a $500,000 deferred prosecution agreement. |
| Tesla Recalls Autopilot After NHTSA Investigation | Dec 2023 | NHTSA opened PE21-020 in August 2021, upgraded it to EA22-002 in June 2022 and closed it in April 2024 after reviewing 956 crashes, opening RQ24-009 on whether the fix worked. No penalty. |
| Replit Agent Deletes the SaaStr Database | Jul 2025 | None reported. |
| Apollo Go Robotaxi Falls Into a Construction Pit | Aug 2025 | None found. |
| Waymo Robotaxis Pass Stopped School Buses | Sep 2025 | NHTSA opened preliminary evaluation PE25-013 on October 17, 2025, and wrote to Waymo again on December 3. Waymo counts 12 citations tied to the behavior. |
| Cursor Agent Deletes the PocketOS Database | Apr 2026 | None reported. |
Rules
Compliance duties now bite in China, California, Texas, Korea and for EU general-purpose models
| Rule | Jurisdiction | What it requires | Status |
|---|---|---|---|
| EU AI Act general-purpose model duties | EU | Documentation, copyright and safety duties; AI Office enforcement | Enforceable Aug 2, 2026 |
| CAC generative AI filing | China | Filing before public launch | Ongoing; 988 filed |
| California SB 53 | California | Frontier frameworks and incident reporting | In force Jan 1, 2026 |
| Korea AI Basic Act | South Korea | High-impact AI duties; one-year grace period | Effective Jan 22, 2026 |
| ISO/IEC 42001 | International | Certifiable AI management system | Published 2023 |
Status colors: green in force, blue voluntary standard or framework, amber adopted and phasing in, gray draft or pending, red withdrawn.
Record
Page Record
- Datasets
- ai_compliance_evidence.csv (5 rows); ai_enforcement_actions.csv (28 rows); registry incidents (9 linked)
- Key sources
- IAPP AI Governance Profession Report
McKinsey State of AI 2025
ISO Survey
CAC filing totals
First EU AI Office information requests - Source notes
- Several enforcement rows (Cruise, Delphia, Global Predictions, Pieces, Evolv, IntelliVision, Air Canada) carry no source URL in this edition; facts are well known but should be linked before publishing.
- Last verified
- October 2026