Quick Buzz Feed
While the legal industry is developing AI tools for administrative and research tasks, these products can still produce 'hallucinations,' necessitating human review at every stage to ensure accuracy and ethical use.
Since ChatGPT's release in November 2022, generative AI has become pervasive, prompting industries, including the legal profession, to consider its adoption. The National Association of Criminal Defense Lawyers (NACDL) emphasizes the ethical responsibility of defense attorneys to integrate AI tools to equalize the playing field, as prosecutors increasingly utilize similar technologies.
Mason Clark's report, 'Parity in Practice: The Defender’s Duty to Ethically Use AI,' advises defense lawyers to prioritize 'enterprise' AI tools like Microsoft CoPilot or legal-specific platforms such as Harvey and CoCounsel over general commercial products due to confidentiality concerns. However, even specialized legal AI tools, including Reuters CoCounsel and Lexis+ AI, have demonstrated a significant rate of generating incorrect information, with studies showing inaccuracy rates of over 17% to 34%.
Clark argues that defense attorneys have an ethical obligation of competence to use AI, especially as opposing counsel adopts these technologies. The report serves as a guide for public defenders and private attorneys to navigate AI products ethically and efficiently, ensuring zealous client defense and preventing negative outcomes that could arise from a failure to utilize available technological advantages. Many firm leaders acknowledge the positive potential of AI while expressing concerns about governing its use.
The president of NACDL, Andrew Birrell, highlights that AI is already influencing courtrooms, citing instances of lawyers presenting AI-generated non-existent legal precedents and AI avatars delivering victim impact statements. This underscores the critical need for defenders to become proficient in AI use to leverage its benefits, such as sifting through vast amounts of evidence or identifying sentencing disparities, while avoiding pitfalls.
The NACDL report classifies AI tasks into three risk categories, each requiring human oversight. Low-risk activities, suitable even for commercial AI tools (though enterprise is preferred), include summarizing general legal principles, drafting administrative documents, and brainstorming. Medium-risk tasks, which affect client matters and need thorough verification, involve drafting initial motions or briefs and research assistance. High-risk applications, such as case-specific factual analysis, strategic decision-making, and courtroom advocacy, should only be supported, not replaced, by AI due to their reliance on privileged information, potential for client harm, and the need for human judgment.
The integration of AI into the legal system is a widespread concern, with various bodies addressing its implications. The Prosecutors’ Center for Excellence supports generative AI for administrative and creative tasks. The American Bar Association advocates for uniform AI standards to address ethical issues like bias and transparency. A Northwestern University survey revealed that over 60% of federal judges use AI, mainly for research and document review, with some formally permitting its use. Proposed Federal Rule of Evidence 707 would mandate reliability tests for AI-generated evidence. Additionally, both the House and Senate have introduced legislation, the Research and Oversight of AI in Courts Act of 2026, to establish a task force studying AI's legal and ethical concerns.