Connecticut Supreme Court Weighs Sanctions Over AI-Generated Fake Citations

The Connecticut Supreme Court is confronting a growing crisis of integrity in legal filings as lawyers submit briefs containing fabricated citations and quotations generated by artificial intelligence. The court is currently addressing instances where legal counsel relied on AI-generated material that lacked any basis in actual law, raising fundamental questions about professional ethics in the age of automation.

According to court records and reports from CT Insider, the court has held sanctions hearings regarding filings submitted by lawyers from the firm GLG Law, including Ian G. Gottlieb. These documents contained "hallucinations"—plausible-sounding but entirely fake legal citations created by large language models. The errors have led to requests from opposing parties for the court to dismiss cases, arguing that the submission of fabricated law undermines the judicial process.

The Rise of AI Hallucinations

This development is part of a systemic risk emerging from the widespread adoption of Large Language Models (LLMs) like ChatGPT. These tools are designed to predict the next likely word in a sequence rather than retrieve verified facts, often resulting in the creation of fictitious case names and citations that appear authentic to an undiscerning eye. This phenomenon has forced judiciaries in various jurisdictions, including Florida, Georgia, and the UK, to implement strict disclosure rules or issue sanctions to ensure human lawyers remain accountable for the accuracy of their work.

Implications for Legal Ethics

The situation in Connecticut represents a critical intersection of emerging technology and legal ethics. When attorneys delegate research to AI without manual verification, they risk compromising the reliability of case law and the overall integrity of the court. The Connecticut Supreme Court's focus on these errors signals that "lax proofreading" is not an acceptable defense for the inclusion of fake citations. The potential for severe procedural consequences, including sanctions, serves as a warning that the duty of competence requires human oversight of all AI-assisted work.

The Path Forward

As courts nationwide increasingly issue warnings against unverified AI material, the legal profession is moving toward a standard of mandatory verification. While the Connecticut Supreme Court continues to weigh the appropriate response to these specific filings, the broader trend suggests a shift toward stricter accountability. Legal practitioners must now navigate a landscape where the efficiency of AI is balanced against the absolute requirement for factual accuracy in court filings.