US Sixth Circuit removes attorney for 'inexcusable' AI hallucinations in appellate filings, setting a judicial benchmark for professional responsibility in AI-assisted legal work
The US Court of Appeals for the Sixth Circuit has removed a lawyer from further representing a client — a man who pleaded guilty to drug trafficking charges — after finding that the attorney had committed 'inexcusable transgressions' by relying on Westlaw's internal CoCounsel artificial intelligence platform for appellate filings without adequately verifying its output. The court found that the lawyer had failed to catch erroneous AI-generated content that appeared in filings submitted to the appellate court. The ruling is significant because it goes beyond the now-familiar pattern of sanctions for AI-hallucinated case citations: it results in actual removal from the case, a more severe professional consequence that directly affects the client's representation. The decision identifies specific conduct — over-reliance on a commercially available legal AI tool without proper supervision — as the trigger for professional discipline, rather than treating the AI error as a purely technical mistake. This is notable because CoCounsel is a product marketed by a major legal publisher (Thomson Reuters/Westlaw) to practising lawyers as an integrated research assistant, meaning the court is explicitly addressing mainstream tool adoption rather than fringe AI experimentation. While this is a US appellate ruling, it carries direct relevance for UK practitioners. The Solicitors Regulation Authority (SRA) has been monitoring AI use in legal practice, and English courts have already seen cases where AI-generated citations caused problems. The professional conduct principles engaged — competence, supervision, and verification — map directly onto SRA outcome obligations and the Bar Standards Board conduct rules applicable to UK lawyers.
Why this matters
This ruling sharpens the professional liability exposure of any lawyer who uses AI tools in court filings without independent verification — a risk that is growing as tools like CoCounsel, Harvey, and Lexis+ AI embed into standard workflows at major firms. Disputes and professional indemnity practices are both activated: for litigation teams, the case defines a supervision standard that must be met; for PI insurers and their counsel, it provides a judicial benchmark for what constitutes negligent AI use. The 'why now' is the rapid mainstream adoption of integrated AI research tools by law firms of all sizes, which has outpaced the development of clear judicial and regulatory guidance. For UK firms, the SRA's ongoing review of AI governance frameworks means this US ruling will be cited in any future UK disciplinary proceeding involving AI-generated content.
On the Ground
A trainee working on any matter involving AI-assisted research would be expected to independently verify every case citation, statutory reference, and factual proposition generated by an AI tool against the original source — this ruling makes that verification step a professional conduct requirement, not a discretionary quality check. On a live filing, they would prepare a verification note flagging every AI-generated assertion and its source confirmation.
Interview prep
Soundbite
Removal from a case, not just sanctions, resets the professional liability bar for AI over-reliance in court filings.
Question you might get
“What professional conduct obligations does this ruling suggest should govern a solicitor's use of AI tools when drafting court documents, and how would you advise a firm to update its AI governance policy in response?”
Full answer
The Sixth Circuit has removed an attorney from a criminal appeal after finding they relied on Westlaw's CoCounsel AI tool without adequately checking its output — producing erroneous content in appellate filings. This is materially more serious than the familiar AI hallucination sanctions cases: the consequence here is removal from the matter, directly harming the client's representation. The ruling establishes that deploying a mainstream commercial legal AI tool without verification constitutes inexcusable professional conduct, not mere technical error. For law firms, this raises the standard of care expected when using AI in any document filed with a court, and for practice management it means AI governance policies must specify mandatory verification steps. In the UK, the SRA is actively monitoring this space, and this ruling will inform both regulatory guidance and any future disciplinary proceedings involving AI-generated legal content.
My notes
saved