AI Regulation News · Weekly Roundup

Weekly AI Enforcement Roundup

The AI rules, rulings, and enforcement actions that changed what a licensed professional owes this week. Primary source on every entry.

Maintained weekly · Last updated July 11, 2026

Every entry below points to a fully reported article on this site. Inclusion follows one test, the same test that governs the tracker. If a development does not change a duty, a liability, or a workflow for a licensed professional, it does not earn a place here. This page keeps the ones that do. It refreshes each week, so you can see what actually moved without reading through more than 260 articles to find it.

This week, that meant courts putting a lawyer's name on unverified AI citations, a regulator in Germany taking an enforcement seat, and a settlement that matters more to practitioners than the ruling behind it. Start where your license lives.

Courts and legal ethics

• A New York appellate panel treated hallucinated AI citations as a sanction, not a slip. The fabricated cites were read as a failure of the lawyer's duty to verify. The rule that binds you is simple: read every citation the model gives you before it reaches a filing. Read the ruling →
• Florida Rule 2.515 puts AI filings on the record. Attorneys in the state now carry a specific sanctions exposure for unverified AI content in what they file. Read the rule →
• The California Bar made verifying AI output an ethics rule. This is no longer guidance a lawyer can weigh. It is a duty. Read the guidance →
• Privilege can survive AI use, if you structure the work right. The Heppner ruling maps where AI-assisted work keeps privilege and where it quietly loses it. Read the analysis →

Healthcare

• The FDA drew the medical-device line through clinical decision support. Cross it, and an AI tool becomes a regulated device, which changes what the clinician who relies on it is answering for. Read the guidance →
• CMS tied an AI patient-safety improvement activity to MIPS. The reporting incentive now reaches how a practice governs its AI. Read the detail →
• Ambient AI scribes are drawing wiretap suits. Consent and recording law has reached the device listening in the exam room. Read the cases →
• Indiana HB1271 targets AI downcoding in record review. Read the bill →

Finance, tax, and insurance

• The first shadow-AI SEC 8-K board disclosure sets a template. Boards have started reporting unmanaged internal AI use as a material risk. Expect others to follow the form. Read the filing →
• FINRA moved on agentic AI and deepfake controls. Read the notice →
• Circular 230 now carries an AI checklist for tax practitioners, reinforced days later by the IRS OPR Alert 2026-19. Read the checklist →
• Maryland HB1563 makes insurers report AI adverse determinations. Read the requirement →
• The CFPB rolled back its fair-lending posture. Most of the duty survived anyway. The headline is deregulation. The reality for a lender is that the obligations that matter still bind. Read what still binds →

Employment and HR

• Four states, four hiring rulebooks, one process you still have to run. The article maps how to keep a single compliant hiring workflow across rules that do not agree. Read the workflow →
• California's CPPA rules now reach automated employment decisions. Read the rule →
• Under California's FEHA, a bias audit is becoming a defense for ad tools. Read the analysis →
• The UK ICO issued guidance on automated decision-making and profiling in recruitment. Read the guidance →

Real estate

• California AB723 forces disclosure of AI-altered listing photos. Read the bill →
• AI tenant screening runs straight into fair-housing disparate impact. Read the analysis →
• The AVM rule now reaches AI home valuation. Read the rule →

Copyright and cross-border enforcement

• The Anthropic copyright settlement, not the ruling, is what practitioners should track. It is the most-read development on this site, and the settlement is where the fair-use line actually gets drawn for people who have to advise on it. The $1.5B Bartz settlement detail carries the numbers. Read the settlement →
• Brazil's PL 2338 advances the Marco Legal da IA. Read the status →
• Germany named its AI enforcer. The Bundesnetzagentur takes the KI-MiG enforcement seat, which tells you who picks up the phone when the EU rules bite. Read the appointment →
• Brazil's ANPD is enforcing on facial recognition and biometrics. Read the action →