From Adam Feldman (Legalytics) yesterday—unsurprising, but it’s helpful that this can be shown concretely:
This article examines what courts have done after lawyers learned, or should have learned, that AI-related inaccuracies had entered the record. The study began with Damien Charlotin’s publicly available AI Hallucination Cases Database and independently reviewed 100 of the most recent cases in the dataset that were resolved U.S. matters involving lawyers. A stricter analytical sample included cases with a meaningful AI connection, a final judicial outcome, and enough information to evaluate counsel’s response after discovery.
The results point to a coherent and practical distinction. The seriousness of the original error remains central: nearly every high-materiality incident produced a serious consequence. But post-discovery conduct also mattered. Serious consequences were substantially more common when lawyers offered misleading explanations, shifted blame, repeated the misconduct, or introduced additional inaccuracies after receiving notice. Prompt admission, by contrast, was associated with a markedly lower rate of serious consequences.
Candor cannot erase significant harm to a client or the judicial process. It can, however, prevent an already serious mistake from becoming a separate problem of professional responsibility….
And some more (read Feldman’s whole post for still more):
The data suggests that the first response should be investigative rather than defensive. Counsel should preserve the prompts, outputs, research histories, drafts, and communications connected to the filing; identify who drafted, reviewed, and signed it; and examine every authority, quotation, and factual assertion produced through the same workflow.
The correction should address the full problem rather than only the error identified by opposing counsel. That may require withdrawing or replacing the filing, notifying the court and opposing counsel, and explaining accurately how the material entered the record. Describing fabricated authorities as typographical errors, or assigning responsibility primarily to a subordinate, client, vendor, or software product, creates additional risk.
Remedial measures are most persuasive when they are concrete and voluntary. Courts credited steps such as mandatory review of primary sources, supervisory approval, AI-use policies, training, audits of related filings, and reimbursement of unnecessary expenses. Once counsel receives credible notice, every later filing, explanation, and omission becomes part of the court’s assessment of professional judgment and candor….
