In a recent incident that brought to light the potential pitfalls of relying on artificial intelligence (AI) in legal proceedings, attorney Steven Schwartz admitted to using ChatGPT, an AI language model, to assist him in drafting a federal court filing. The AI-generated document included citations to nonexistent cases, causing embarrassment and raising concerns about the reliability of AI-generated content. As a result, Judge Brantley Starr has implemented new rules in his courtroom, requiring attorneys to confirm that no part of their filing was drafted by generative AI or to ensure that AI-generated language has been verified by a human. This development highlights the need for caution when using AI in legal research and underscores the importance of human oversight in the legal profession.
Judge Starr’s Requirement for Certification
Judge Brantley Starr, a Texas federal judge, has introduced a mandatory certification for attorneys appearing in his court. Attorneys must now file a certificate on the docket, declaring that no part of their filing was drafted by generative AI, such as ChatGPT, or affirming that any AI-generated language has been reviewed for accuracy by a human using reliable legal sources. This certification requirement aims to prevent the submission of unreliable or fabricated information in court proceedings.
Reasoning Behind the Certification Requirement
The memorandum explaining the necessity of the certification highlights the limitations and potential risks associated with AI-generated content. The following points were made:
- AI platforms like ChatGPT have their uses in the legal field, such as generating form divorces, discovery requests, or anticipated questions at oral arguments. However, they are not suitable for legal briefings, as they are prone to hallucinations and bias.
- AI language models can produce false information, including made-up quotes and citations, compromising the integrity and accuracy of legal arguments.
- Unlike human attorneys who swear an oath to uphold the law and set aside personal biases, AI systems lack a sense of duty, honor, or justice. They act according to programmed algorithms, not guided by principles or legal ethics.
- The reliability and accuracy of AI platforms for legal briefing remain uncertain. Any party believing an AI platform has the necessary accuracy and reliability must justify its use.
The Consequences of Misusing AI in Court Filings
The case involving Steven Schwartz and his reliance on ChatGPT as a legal research source serves as a cautionary tale. The repercussions of his actions may include potential sanctions or disbarment for both Schwartz and his legal partner, Paul LoDuca. The court has set a hearing for June 8 to address this unprecedented circumstance and allow the attorneys an opportunity to explain their actions.
Lessons Learned and Future Implications
The incident involving AI-generated content in a court filing underscores the need for prudence and verification when utilizing AI in legal research. Some key takeaways include:
- AI language models can be a powerful tool but must be used judiciously and with proper oversight.
- Legal professionals should exercise caution when relying on AI-generated content and conduct thorough verification of its authenticity and accuracy.
- Human oversight and critical evaluation remain essential to ensure the integrity and reliability of legal arguments.
- Judge Starr’s certification requirement may serve as a model for other courts seeking to address the challenges posed by AI-generated content in legal proceedings.
While AI can be a valuable asset in various domains, its limitations and potential risks must be recognized and addressed to maintain the credibility and fairness of legal processes.