Criminal defense lawyers have duty to use AI, paper argues

In Brief

• Association’s new white paper argues criminal defense lawyers have an ethical duty to understand and responsibly use AI tools.
• Prosecutors already use AI for evidence management, discovery review and conviction reassessments, creating concerns about resource disparities.
• Defender organizations across several states report successful AI use in reviewing bodycam footage, identifying inconsistencies and improving case management.
• The report recommends formal AI policies, staff training, vendor reviews and strict verification of AI-generated content to protect clients’ rights.

In just a few years, attorneys have gone from being instructed to be wary of AI to being told to become proficient in AI. A new National Association of Criminal Defense Lawyers (NACDL) paper argues that defenders have an ethical duty to use AI.

“Every day, our members stand between individual clients and a government with vastly superior resources regarding investigators, data, and time,” said Lisa M. Wayne, executive director of the Washington, D.C.-based NACDL. “No single defender can match that. Used carefully and ethically, AI is one of the few tools that can help level that field, and we have a duty to understand it well enough to use it for our clients without ever compromising their rights.”

Generative AI has entered legal practice at an extraordinary pace. Prosecutors’ offices across the country are already using AI to manage evidence, streamline discovery, and review old convictions for possible resentencing.

These tools are rapidly becoming part of legal research, document review, investigations, and trial preparation, often outpacing ethical guidance. NACDL’s white paper proposes guidelines for defense lawyers to use these tools responsibly while remaining competent, preserving confidentiality, and maintaining candor with the court.

“We put this out now because the technology is not waiting for us,” said Andrew Birrell, president of the National Association of Criminal Defense Lawyers. “In five years, asking whether a defender uses AI will sound like asking whether they use a computer. It will just be the job.”

“Prosecutors are using it to manage evidence and speed up review. If the defense sits this out, we do not stay neutral. We fall behind. And when the defense falls behind, the client pays for it,” Birrell asserted.

“The danger was never that defenders would use AI,” Birrell added. “The danger is that only one side does.”

“The criminal legal system operates on adversarial principles,” the authors write in the paper. Prosecutors and defense lawyers each advocate zealously for their respective positions, and the truth-seeking function depends on both sides presenting their strongest cases. When one side gains significant technological advantages, this balance shifts.”

Uses of AI have included reviewing bodycam footage, finding testimony inconsistencies, and reducing manual data entry by up to 85%.

The report highlights defender offices using AI to improve casework:

• The Kentucky Department of Public Advocacy uses JusticeText to quickly review bodycam footage.
• The California Innocence Project has used AI to help uncover potential wrongful-conviction issues and prioritize cases for attorney review.
• The Maryland Office of the Public Defender uses eDefender to securely centralize case files, tasks, and communications for defense teams.
• The Colorado State Public Defender’s Office uses Reduct to highlight key clips, generate captions, and create court exhibits from bodycam footage and police interrogations.
• A Wisconsin defender used JusticeText to uncover evidence of police misconduct more efficiently.

“Used well, AI is the closest thing an under-resourced defender has ever had to a force multiplier. It lets one lawyer do the review that used to take a team,” Birrell emphasized.

“With the right guidance and the right guardrails, generative AI can genuinely improve how defenders practice: sharpening legal research, lightening workloads, and freeing lawyers to spend more of their time on the judgment calls that only a human advocate can make,” said Mason R. Clark, the report’s lead author and an assistant professor of law at St. Mary’s University School of Law in San Antonio, Texas. “That’s the opportunity here, and this white paper is meant to help defenders seize it effectively.”

The authors recommend that defense lawyers and offices adopt formal AI use policies within one year, while recognizing that implementation timelines may vary by office and system.

NACDL’s guidance promotes responsible AI adoption to help the defense bar narrow resource disparities while upholding the ethical standards of zealous advocacy. The paper also provides guidance on vendor review, risk assessment, staff training, AI policies, and a model policy template.

The authors also are aware of and open about the risks of AI. In fact, the paper was drafted with AI assistance, and then a human verified every single citation by hand. “ That is exactly the method we are asking defenders to use, so we held ourselves to it first.”

“[T]reat AI output like the work of a sharp new associate who has never set foot in a courtroom,” Birrell said. “You check the citations. You read the cases. The tool drafts. The lawyer decides. That line does not move.”

The rise of AI-generated evidence requires defenders to be alert to the possibility of altered or fabricated materials entering a case from any source, the authors note. To protect clients, defenders should closely scrutinize digital evidence, verify AI-generated content, and use forensic experts when necessary to evaluate its authenticity.

“That’s exactly why we can’t just tell defenders to go use AI,” Clark said. “We must tell them how: How to verify what it gives them, how to protect client confidentiality, and how to keep their own judgment squarely in the driver’s seat.”

While lawyers are recognizing the benefits of AI, the public may still be wary about it. “A defender should be prepared to explain how forgoing AI could be beneficial or detrimental to the case, especially if opposing counsel is likely to use it,” the authors say.

The white paper is called “Parity in Practice: The Defender’s Duty to Ethically Use AI.” A version of this will be available in the inaugural edition of The Journal of Advocacy and Litigation.

Following the release of the white paper, NACDL’s Task Force on Artificial Intelligence will provide ongoing training to help defenders integrate AI into their practices. The task force will also develop and share model policies and practical guidance for the responsible use of AI. These resources are designed to support defenders at every budget level, from solo practitioners to statewide public defender systems.

“This isn’t a one-and-done report; it’s the start of a dialogue we intend to keep having with our members,” said Wayne. “Technology isn’t going to slow down, and neither will we. I want every defender in this country walking into a courtroom with the sharpest tools available and the know-how to use them well, and that’s exactly what we intend to deliver.”

“The judgment, the investigation, the call with a scared client at 2 in the morning, none of that is going anywhere, and none of it should,” Birrell emphasized. “ What AI takes off your plate is the grind. The document review. The first-draft timeline. Do that faster and you have more hours for the human work, not fewer.”