In 2026, 50% of Americans reported that they had utilized AI chatbots, a dramatic increase from the 33% of chatbot usage reported in 2024. That shift is already being felt inside courtrooms nationwide. A Mississippi federal judge recently found herself sorting through AI-corrupted filings from both sides of a case, as the attorneys submitted briefs riddled with AI-generated errors. Congressional oversight letters reveal that two federal district judges — one in New Jersey and one in Mississippi — issued decisions containing serious factual inaccuracies linked to apparent AI use in their chambers.  Even a self-described “AI expert” was caught submitting an AI-generated declaration containing fabricated case citations. It would be naive to assume that generative AI has not already found its way into the jury room.

In the internet era, courts have long grappled with juror misconduct – a juror Googling a definition, reading a news article, or searching for a party on social media. Juror usage of generative AI is far more threatening to trial integrity than traditional internet research. When a juror conducts a search on google and lands on a Wikipedia article, they likely recognize what they are reading as informal and unverified. However, when a juror asks generative AI the standard for negligence in a personal injury case, they receive a confident, well-structured, legal-sounding answer with citations. That AI-generated answer will not account for whether the case is civil or criminal, federal or state, or which state’s law governs. It may also contain hallucinations including fabricated statutes, invented case citations, or invented legal rules. None of that will be apparent to a common juror. The result is a juror who believes they are better informed, when in fact they have received false legal instruction from a source no judge authorized and no attorney can cross-examine.

The existing prohibition is clear: jurors must decide cases based solely on evidence admitted at trial and instructions given by the judge. However, courts nationwide face many challenges in enforcing this standard. The court must first determine whether there was juror misconduct. If juror misconduct is found, the court must then determine if the misconduct actually influenced the verdict. When the alleged misconduct involves the use of AI, it is particularly challenging to determine the AI chatbot utilized, what exactly the juror searched, the actual answer they received, and then how it impacted the deliberations.

There has not yet been a report of a civil verdict being overturned due to AI juror misconduct. However, this does not mean that there has not been juror misconduct through the use of generative AI, as it likely went undetected. Defense counsel cannot afford to wait for documented instances of juror misconduct through generative AI. Proactive trial management is the only safeguard while courts, bar associations, and legislatures work to catch up. When preparing for trial, counsel should consider the following steps:

• Updated voir dire: Ask prospective jurors directly about their use of generative AI and whether they know those tools sometimes provide false information.
• ​Demand explicit AI instructions: Request jury instructions that specifically prohibit consultation of generative AI by name — not just generic prohibitions on “internet research.”
• ​Vet opposing counsel’s filings: AI-generated hallucinations in briefs and motions are now common enough that defense counsel should verify opposing citations as a matter of course.
• ​Scrutinize expert work product: All expert witnesses should be questioned about their AI tool use in preparing reports and opinions.

•​Preserve the record for post-trial relief: If juror misconduct through the use of AIis suspected, investigation of juror conduct is time-sensitive and subject to juror secrecy rules that vary by jurisdiction. Counsel who have built a trial record on AI through voir dire, jury instructions, and preserved objections are better positioned to support a motion for new trial.

To learn more about how CMBG3’s Trial Division can assist you, please contact Veronica Lee at vl**@***g3.com.