The US Court of Appeals for the Fifth Circuit on Tuesday became the first US appeals court to propose a new rule requiring lawyers to certify that they either did not use generative artificial intelligence (AI) programs, like Chat GPT, to draft filings or that humans reviewed AI-generated material. The proposal aims to ensure that the use of AI in legal work is transparent and accountable.

According to a report by law firm software company Clio, AI can offer speed, efficiency and cost-effectiveness benefits. Still, it also poses risks such as errors, ethical and regulatory concerns, and algorithm bias.

In February, law firm Allen & Overy integrated Harvey, an AI platform specializing in machine learning and data analytics, into its international practice. About 3,500 lawyers asked Harvey 40,000 questions about their day-to-day work during a trial period.

However, in the context of lawyers using artificial intelligence, AI poses problems such as “hallucinations,” unpredictability, and response divergence. Hallucinations are incorrect outputs that could lead to tort liabilities, consumer harm or regulatory breaches. Unpredictability arises due to a lack of transparency, making it difficult to check whether the model meets standards of quality and accountability. Response divergence is another issue due to the nature of AI models, which can give multiple answers to the same question. To mitigate these risks for the law, researchers have proposed explainable artificial intelligence (XAI) models that can provide reasoning behind AI-generated predictions or recommendations to allow users to discern whether AI is “right for the right reasons.”

After banning the use of AI in his courtroom in June, Judge Brantley D. Starr of the US District Court for the Northern District of Texas said, “While attorneys swear an oath to set aside their prejudices, biases, and beliefs to faithfully uphold the law and represent their clients, generative artificial intelligence is the product of programming devised by humans who did not have to swear such an oath.”  Another AI-rules proposal from that court states “although technology can be helpful, it is never a replacement for abstract thought and problem-solving.”

According to the new rule, lawyers who misrepresent their compliance with the court could have their filings stricken and sanctions imposed against them. The 5th Circuit accepts public comment on the proposal rule through January 4.