AI & the Law: A Federal Courts Newsletter - October 22, 2026
U.S. Courts sent this bulletin at 10/22/2026 01:59 PM EDTCoPilot for Legal, ChatBot Regs, & AI in the 5th Circuit
Having trouble viewing this email? View it as a Web page.
This news service and the companion Artificial Intelligence (AI) Resources site were created by the AO's Library Technology Working Group. To access these articles remotely or from a mobile device, you may need to be connected to the Judiciary network (VPN or DCN). Contact your local court librarian if you have issues accessing full articles.
Anyone With a Subpoena Can Read Your AI Chats. That’s a Risk
US Law Week | October 22, 2026
The tools the state uses to follow the money — subpoena, warrant, wiretap, know-your-customer rule — focus on a person. An investigation starts with a suspected crime; it works back to the person responsible by compelling the banks and other parties who dealt with him to turn over evidence in their possession. The whole system rests on having someone to compel.
Software agents take that someone out of the transaction. When an agent’s payments draw scrutiny, the question will be what the human authorized. Answering that question means reading the instructions as well as all the deliberation behind them: the drafts, doubts, and iterative queries.
Illinois Judges are Using AI Tools, Misuse Scenarios Yet to be Tested
Law.com | October 22, 2026
Illinois judges are increasingly incorporating artificial intelligence into legal research and other courtroom work, encouraged by tools that can accomplish tasks within minutes, rather than hours. But as AI becomes embedded in judicial work, Illinois has yet to publicly test what happens when a judge’s use of the technology goes wrong.
The Illinois Supreme Court’s AI policy, which was adopted in January 2025, allows litigators and judges to use the technology while emphasizing that legal professionals remain responsible for their final work product.
Anthropic’s Ambitions in Legal Technology Get a Reality Check
US Law Week | October 22, 2026
Law-firm innovation leaders attended a private meeting with Anthropic's new head of Claude for Legal, where they were told the company won't negotiate key contract terms around privacy and data security unless they commit to a minimum amount of spending.
Should Legal Tech Vendors Fear Microsoft's Copilot for Legal?
LegalTech News | October 21, 2026
Experts suggest that Copilot for legal is likely to achieve broad adoption and could pose a threat to point solutions that don’t differentiate from general-purpose AI.
5th Circuit Clerk Retires After Helping Chart Court's AI Adoption
National Law Journal | October 20, 2026
“I spent much of my last few years on AI policy for the courts, and I don’t intend to walk away from it,” Lyle Cayce said. “The questions courts face are changing fast, and those of us who worked through the early policies have a responsibility to share what we learned.”
Del. Ruling Shows How AI Meeting Transcripts Can Backfire
Law360 | October 20, 2026
On Aug. 28, Vice Chancellor Lori W. Will issued a post-trial opinion in ATG Capital Opportunities Fund LP v. Lane, holding that the board of Empery Digital Inc. improperly rejected an activist investor's director nomination notice and, in so doing, breached its fiduciary duties.
In reaching its decision, the Delaware Court of Chancery considered not only official board minutes but also transcripts generated by artificial intelligence from the same board meetings. The two records of the same meetings differed, and the court relied on the AI-generated transcripts to supply additional context that was not present in the official minutes about the board's objectives and decision-making process.
AI Misuse Is a Serious Problem for Lawyers
Legal Intelligencer | October 20, 2026
Lawyer AI abuse has been reported worldwide in over 2,000 legal matters. What should a lawyer do?
NYC Proposes First-of-Its-Kind Automated Customer Service Chatbot Regulation
New York Law Journal | October 20, 2026
The New York City Department of Consumer and Worker Protection (DCWF) and commissioner Samuel A.A. Levine announced the proposal, with Levine saying the rule is the next step after the department previously banned subscription traps, known as its "Click to Cancel" rule.
The chatbot rule would require businesses to provide timely and reasonable customer service in addition to offering customers access to a human representative if that service falls short. The proposal builds off the city's consumer protection law and includes fines as ranging from $525 to $3,500. If passed, New York would be the first municipality in the nation to enforce this level of consumer protection.
OpenAI Confronts Deluge of Social Media-Style Safety Lawsuits
US Law Week | October 19, 2026
OpenAI Inc. is confronting a new wave of litigation from state attorneys general and plaintiffs’ law firms who are drawing on the lessons of recent, successful social media harm cases to advance consumer lawsuits against the AI lab.
The San Francisco company was sued by over half a dozen state attorneys general last week alleging it failed to implement sufficient safeguards on ChatGPT to protect vulnerable users including children.
Judges See Promise, Peril as AI Moves Into the Courts
American Lawyer | October 15, 2026
Artificial intelligence is being implemented just about everywhere, including in the courthouse. While recently podcasting on behalf of Legal Speak, I had the opportunity to interview judges at RelFest in Chicago about their take on AI's growing role in the courts.
The conversations revealed that some jurists are focused on AI's potential to make the justice system more efficient and accessible, while others are concerned about whether courts are prepared for increasingly sophisticated AI-generated evidence.
What No One Is Talking About in Connecticut’s Prompt Injection Ruling
Texas Lawyer | October 15, 2026
Matthew Elliott, a self-represented plaintiff, embedded three-point white font against a white background inside his motions. Invisible to a human reader, the hidden text contained instructions for any large language model that might ingest the document, telling the system to agree with his filing and treat an earlier clerk's ruling as an error. In later filings, Elliott added additional hidden phrases, including a greeting and a link to a video. He characterized the original injection as an "audit" of the court's systems and the later additions as jokes, explanations the court rejected as implausible and found to be deliberate misconduct.
The attempt failed. Connecticut state courts do not use artificial intelligence to review filings, so no system or clerk was misled. However, as the court noted, the attempt was far from aimless. Opposing counsel routinely screen incoming filings with AI tools, and the hidden instructions were aimed directly at those automated readers, including defense counsel in this case.
Scientists are Finding AI May Make Us Remember Things That Never Happened
Washington Post | October 15, 2026
Studies show that artificial intelligence can distort memories, raising questions about what happens when machines become keepers of the past.