Connecticut's appellate and supreme courts publish new generative AI rules imposing immediate sanction liability on filers for AI-generated errors

The Connecticut Appellate Court and Connecticut Supreme Court have published new rules on the use of generative AI (artificial intelligence tools that produce text, citations, or arguments from large datasets) in court filings, which took immediate effect. The rules outline additional pathways for courts to impose sanctions on filers where AI-generated errors appear in submitted documents, placing responsibility squarely on the attorney or party who filed the document rather than on the AI tool itself. The move follows a pattern seen in other US jurisdictions where courts have penalised self-represented litigants and, in some cases, qualified lawyers for submitting filings that contained AI-hallucinated case citations (fictitious or inaccurate case references generated by AI systems). While Connecticut is a US state jurisdiction, the regulatory direction is directly relevant to English-law practitioners because the UK courts and professional regulators including the Solicitors Regulation Authority (SRA) are actively monitoring AI use in litigation and the SRA has already referred a solicitor to a disciplinary tribunal in an unrelated matter this week, illustrating the heightened scrutiny on solicitor conduct more broadly. The core legal issue crystallised by the Connecticut rules is verification responsibility: the lawyer signing the document bears full professional liability for its accuracy, regardless of the tool used to draft it.