AI hallucinations cost lawyer $15,000 in fines | The Business Standard
Can artificial intelligence draft legal documents? Certainly. Can it do so without fabricating cases? That remains debatable.
A federal magistrate judge has recommended a $15,000 sanction against Texas-based attorney Rafael Ramirez after he cited fictitious court cases generated by an AI tool. As The Register reports, Judge Mark J Dinsmore of the Southern District of Indiana found Ramirez guilty of failing to verify the authenticity of three cases in his legal brief. Each citation, it turns out, was entirely fabricated. The penalty, set at $5,000 per erroneous brief, serves as a reminder that legal professionals must uphold accuracy, even when relying on artificial intelligence.
Ramirez admitted during a January hearing that he had previously used AI tools for drafting agreements and other legal matters. However, he was unaware of their tendency to generate false citations. He did not act in bad faith, Ramirez argued, but conceded his failure to comply with Federal Rule of Civil Procedure 11, which requires attorneys to ensure the accuracy of their submissions.
The concern about AI hallucinations
The case underscores a growing concern about AI hallucinations — where generative models produce plausible but entirely false information. Similar incidents have surfaced, including one involving Minnesota Attorney General Keith Ellison. In defending a state law on deepfakes, Ellison submitted an expert report containing non-existent academic references generated by GPT-4o.
