Defense attorneys urged to cautiously adopt AI to match prosecution

WASHINGTON (CN) — Since the release of ChatGPT in November 2022, generative artificial intelligence products have become more widespread throughout the economy, shifting the focus of countless industries to adopt the technology while raising questions regarding its use throughout the government and the nation’s institutions.The legal profession is no different, the National Association of Criminal Defense Lawyers argue in a report released Thursday. The group says defense lawyers have the “ethical obligation” to responsibly adopt certain AI tools and level the playing field as prosecutors increasingly utilize such products.In the report, “Parity in Practice: The Defender’s Duty to Ethically Use AI,” author Mason Clark, assistant professor of law at St. Mary’s School of Law, lays out a set of opportunities and risks involved with AI tools in their current state.Rather than adopt commercial AI products such as OpenAI’s ChatGPT, Google’s Gemini or Anthropic’s Claude, each of which carries significant confidentiality concerns, defenders should consider and vet “enterprise” AI tools such as Microsoft CoPilot via a Microsoft license or legal-specific tools such as Harvey and CoCounsel.While built around the legal profession and its ethics code, the law-specific tools still suffer from the same issues commercial tools, such as the Reuters CoCounsel tool powered by OpenAI’s GPT-4 technology, and the LexisNexis tool Lexis+ AI.According to a study by the Stanford RegLab and Institute for Human-Centered Artificial Intelligence, the Lexis+ AI and Ask Practical Law AI systems produced incorrect information over 17% of the time, while Westlaw, into which Reuters integrated CoCounsel for AI-Assisted Research, hallucinated more than 34% of the time.In an interview with Courthouse News, Clark explained the report was intended to provide a roadmap for defense attorneys and public defenders to understand how they can avoid common pitfalls associated with AI products and utilize them in an efficient and ethical way to “zealously defend” their clients.“I went into it thinking this paper needs to be about a call to action for defenders to ethically use this to balance the scales in the criminal legal system, and although there may be skepticism, although there may be hesitance, we have an ethical duty of competence to use the technologies that the adversarial opponents are also using,” Clark said.He noted that, when speaking with defense attorneys for the report, many firm leaders expressed a positive outlook on using AI tinged with a sense of concern regarding their attorneys’ use of personal tools, a “black box” that’s difficult to uniformly govern, compared to prosecutors’ more universal and enterprise-level tools.“Some prosecutors’ offices are increasingly experimenting with enterprise-level AI tools, though resource disparities may hinder comparable adoption among some defender organizations,” Clark wrote in the report. “Moreover, AI has already impacted workflows and organizational structures at firms and defenders’ offices, and its ubiquity encourages skill flattening. What is clear, though, is that defenders have an ethical obligation to use AI in their practice and in the courtroom, and a failure to do so could produce incompetent representation of and negative outcomes for criminal defendants.”That obligation is separate from already infamous examples of AI creeping into courthouses across the country, from lawyers citing nonexistent legal precedents hallucinated by their AI chatbots to victim impact statements from the deceased victim themselves created and read by an AI avatar, association president Andrew Birrell said in a statement.“AI is already inside the courtroom,” Birrell said. “The defenders who learn to use AI well are the ones who are going to out-lawyer the other side: catching the contradiction buried in 10 hours of bodycam footage, or the sentencing disparity nobody had time to find by hand. That’s a real opportunity to do right by our clients, and defenders have an ethical duty to rise to it.”According to the association, there are several tasks AI could be used for that it deemed low risk, medium risk and high risk, all of which still require some level of human input and review.Low-risk activities include anything unlikely to produce errors, such as having an AI tool summarize general legal principles unrelated to a specific case, drafting administrative documents such as office policies or internal memos, generating routine correspondence templates, creating educational training materials and brainstorming approaches to common legal issues. Such tasks could be accomplished with commercial AI tools, but the association still recommended enterprise-level systems.Medium-risk tasks include any work that affects client matters that can be thoroughly verified beforehand, such as drafting initial versions of motions or briefs, research assistance or summarizing long documents, again with human oversight.According to the Vanderbilt Project on Prosecution Policy, prosecutors have found AI tools promising for “litigation content” like draft jury selection questions, opening statements, direct examination questions, motion and appellate arguments.Finally, high-risk applications include any task involving the use of privileged and/or confidential information and those where errors could directly harm clients or requires human judgement, strategy or analysis. Case-specific factual analysis, strategic decision-making, client counseling and courtroom advocacy are all considered high-risk by the association, and thus AI should only support, rather than supplant, human judgment.All parties involved in legal disputes have moved to address the use of AI tools in courthouses across the country.The Prosecutors’ Center for Excellence, a nonprofit group focused on providing best practices for prosecutors, argued in a January 2025 report that generative AI tools can provide prosecutors a leg up their cases by assisting with many administrative, creative and training purposes.The American Bar Association has also pushed for uniformity regarding AI standards, urging courts and lawyers to address the emerging ethical and legal issues, including bias and transparency in automated decisions.According to a March 2026 survey conducted by Northwestern University of 502 randomly selected federal judges, over 60% of respondents said they use at least one AI tool in their chambers, mostly for legal research and document review. Further, about 25% formally permit the use of AI in their chambers, while 20% ban it.The U.S. Judicial Conference’s Committee on Rules of Practice and Procedure recently proposed Federal Rule of Evidence 707, which would subject any AI-generated evidence offered at trial without a corresponding human expert to a reliability test.In February, U.S. District Judge Jed Rakoff in the Southern District of New York determined that a criminal defendant’s unsupervised communication with an AI chatbot was neither protected by the attorney-client privilege or the work product doctrine. However, the issue left open questions as to whether such doctrines apply when an attorney directs such AI usage.In March, both the House and Senate introduced the Research and Oversight of AI in Courts Act of 2026, which would establish a task force to study legal and ethical concerns regarding AI transcription and speech recognition technologies. The bills are currently pending before their respective Judiciary Committees.Subscribe to our free newslettersOur weekly newsletter Closing Arguments offers the latest about ongoing trials, major litigation and rulings in courthouses around the U.S. and the world, while the monthly Under the Lights dishes the legal dirt from Hollywood, sports, Big Tech and the arts.Additional Reads