A federal judge sanctioned a law firm for submitting AI-generated briefs citing fake cases.
The sanctions didn’t just hit the associate who ran the research through ChatGPT without verifying it. The judge held senior partners personally liable for the errors, requiring them to pay fines from their own pockets because they failed to supervise AI use on client matters.
Thomson Reuters issued a public statement clarifying that their AI legal tools are designed to support attorney judgment, not replace it, and that liability for final work product belongs entirely to the human attorney using the tools. The message was clear – AI vendors won’t indemnify lawyers for hallucinations and mistakes.
Courts have moved past gentle warnings. Monetary fines, public reprimands, and state bar referrals are now standard responses to AI-generated filings that cite cases that don’t exist. Over 1,300 documented cases worldwide involve AI hallucinations in legal work.
Why Partners Are Now Personally Liable
Law firm partners have always had supervisory responsibilities for associates’ work, but AI creates new accountability.
When an associate generates a brief using AI and the partner signs it without adequate verification, the partner is personally responsible for the AI’s errors. Courts aren’t accepting “I didn’t know the associate used AI” as a defense.
Partners who approve filings containing fake cases face sanctions even when they didn’t personally use the AI tool. The supervisory duty includes ensuring adequate verification processes exist and are followed before work product goes to court.
Some judges have made partners pay sanctions from personal funds rather than firm accounts to emphasize individual accountability. The penalties are designed to make partners take AI oversight seriously at a personal financial level.
What This Means for Outside Counsel Relationships
General Counsel managing outside counsel relationships need to understand their law firms are using AI on client matters.
The firms may not volunteer this information. They’re using AI for research, drafting, and document review because it increases efficiency and profitability. But if their oversight processes are weak, the exposure doesn’t stay contained at the firm.
Your company’s name is on court filings. Your litigation strategy is in those briefs. Your outcomes are affected when a judge discovers fake citations and loses patience with your counsel. The client suffers reputational damage and strategic disadvantage even though the firm made the mistake.
Bad AI work product from outside counsel can derail cases, create adverse precedent, and damage relationships with judges that affect future matters.
The Right Questions to Ask Law Firms
Asking outside counsel, “do you use AI” is no longer sufficient. Most firms use AI in some capacity, and many won’t admit it due to client concerns.
The better questions focus on oversight and accountability. What verification process exists for AI-generated work product? Who inside the firm owns accountability for ensuring AI outputs are accurate before filing? What training have attorneys received on AI tool limitations and verification requirements?
Ask about specific cases – if an associate uses AI for research, what review happens before that research goes into a brief? If AI drafts documents, what human review confirms accuracy? How does the firm ensure AI hallucinations don’t make it into client work?
Request the firm’s written AI use policy if one exists. Many firms are still operating without formal policies, which is a red flag for inadequate oversight.
The Verification Problem
The fundamental issue is that verification takes time, which eliminates AI’s efficiency advantage.
If an associate uses AI to research and draft a brief in two hours, but verification requires a partner to independently check every citation and legal conclusion, the time savings disappear. The partner might as well have done the research traditionally.
Some firms are cutting corners on verification to preserve AI efficiency gains. They’re treating AI output as reliable and only spot-checking rather than thoroughly verifying. This is where hallucinations slip through and create sanctions.
Proper AI oversight requires assuming AI output may be wrong and verifying everything critical before filing. Few firms have implemented this level of rigor consistently.
AI Vendor Disclaimers
Legal AI vendors uniformly disclaim liability for errors in their tools’ outputs.
Terms of service state that AI is provided “as is” for informational purposes, users are responsible for verifying accuracy before relying on outputs, and the vendor assumes no liability for hallucinations or incorrect legal analysis.
Thomson Reuters’ statement made this explicit after lawyers tried arguing the AI vendor should share liability for fake cases generated by their tools. The vendor’s position is clear – attorneys are professionals with judgment obligations that can’t be outsourced to algorithms.
This means law firms bear full liability for AI errors regardless of which vendor’s tools created the mistakes.
State Bar Discipline
Beyond court sanctions, attorneys face state bar discipline for AI-related misconduct.
Bar associations are investigating lawyers who submitted fake AI-generated cases. Discipline ranges from private reprimands to public censure to suspension, depending on severity and whether the attorney tried to cover up the mistake after discovery.
Some bar complaints allege violations of competence requirements – attorneys must understand the tools they use, and using AI without understanding its limitations and verification needs constitutes incompetence.
The professional consequences extend beyond individual case sanctions and can affect an attorney’s ability to practice.
What General Counsel Should Do
GCs should have explicit conversations with outside counsel about AI use and oversight before problems emerge in live matters.
Make clear that AI use is acceptable but requires rigorous verification and that client expectations include human judgment at every stage. Request notification if AI will be used on critical filings or high-stakes matters.
Consider including AI oversight requirements in engagement letters, specifying verification standards and supervisor accountability for AI-generated work product.
Monitor court filings from outside counsel for signs of AI use without adequate review – unusual citation patterns, formatting inconsistencies, or legal arguments that seem off-base may indicate AI drafting without proper verification.
If sanctions or errors occur, evaluate whether the firm’s AI practices created systemic risk across all your matters with them, not just the one where problems surfaced.
The Stakes
AI in legal work is only as safe as the oversight structure around it.
Firms with strong verification processes and clear accountability can use AI to increase efficiency without creating risk. Firms treating AI as a shortcut without adequate human review are creating liability for themselves and their clients.
The 1,300 documented cases of AI hallucinations represent lawyers who didn’t verify AI outputs before filing them. Courts are making examples of these cases to send a message – technology doesn’t excuse professional responsibilities.
General Counsel who assume their outside counsel handles AI appropriately are taking unnecessary risks with their company’s litigation interests.
Brian Alexander
Founding Partner, VisionAI+ Consulting Group
Former General Counsel, AI Engineering Firm
Certified ISO 42001 Lead Auditor