Jay Baumgardner, a Muso.AI co-founder and opposing litigant in the Delaware case, at a recording console. Source: Advanced Alternative Media.

There’s another fight over a litigant’s AI conversations. The co-founder of a music company is asking a court in Delaware to protect his Claude chats. The underlying dispute is about control of the business, Muso.AI. His opponents requested his “complete Claude history” concerning the company, the lawsuit and several related subjects. He responded with a motion for a protective order, arguing that conversations used to prepare his case reveal his legal strategy. He has represented himself for much of the litigation and says those prompts and drafts should receive work-product protection. As of October 26, there was no ruling. The dispute is reminiscent of Shmuley Boteach’s fight in Florida, which we covered earlier this month.

Courts

Argentina’s Palace of Justice in Buenos Aires. Photo: ProtoplasmaKid / Wikimedia Commons, CC BY-SA 4.0 (cropped).

Argentina’s national judiciary plans to use AI to transcribe court hearings without sending recordings to outside providers. The service, scheduled to become available Wednesday, will turn uploaded audio and video into searchable text. The judicial council announced Friday that all processing will run inside the judiciary’s own data center.

American courts face a shortage of court reporters, which we covered in June. California’s judicial branch reported in July that more than three million hearings in family, probate and certain civil cases proceeded without a verbatim record between April 2023 and March 2026. The judicial branch says missing records can make it much harder for litigants to appeal.

Business

Delaware Legislative Hall in Dover. Photo: w_lemay / Wikimedia Commons, CC BY-SA 2.0 (cropped).

Delaware’s plan to let AI run companies is drawing congressional scrutiny. The proposed “agent-managed company” could enter contracts, own property and sue or be sued under an AI agent’s direction. Its human or corporate owner would generally be protected from its debts, provided it adequately funds the company and doesn’t use it for fraud or deliberate lawbreaking. Delaware proposes a 30-month supervised test, with officials approving participants and requiring capital to cover potential claims.

House Oversight Chairman James Comer requested records Monday, questioning whether the structure would shield people from responsibility for harmful AI actions. He also questioned industry involvement, including Delaware’s partnership with legal-technology company Norm AI. The proposal’s developers argue that giving AI-run businesses a legal identity would make their actions traceable and provide a target for damages. Officials could suspend participants and seek their dissolution. The proposal hasn’t become law; the governor’s office said Monday it was reviewing Comer’s letter.

In Other News

Amid a broader surveillance debate, the sheriff’s office in Ulster County, New York, rejected AI report writing after a 90-day trial. (Kingston Wire; Daily Freeman)

A local news report says prosecutors in Utah and Texas are using AI to review evidence. (KSL)

Chicago is having its own license-plate-reader reckoning, investigating possible misuse and tightening officers’ access to outside systems. (WTTW)

Akron, Ohio, blocked out-of-state police from directly searching its Flock license-plate records. (Ideastream)

New York police are negotiating a $750,500 purchase of portable devices that mimic cell towers to locate phones. (NYC procurement notice)

A New Jersey Assembly committee advanced a bill to bar arrest warrants based solely on facial-recognition matches. (NJ State House News Service)

A Colorado federal judge said keeping AI CSAM at home can be a crime, disagreeing with the Seventh Circuit. (Court order)

Pennsylvania may replace county-by-county AI filing rules with a statewide accuracy certification, while its new AI advertising law leaves uncertainty over when disclosure is required. (Pennsylvania Bulletin; Loeb & Loeb)

The U.S. cybersecurity agency confirmed exploitation of flaws in PHP, software widely used to run websites.

Virginia ordered a statewide plan to prevent and respond to AI cyberattacks. (Governor’s office)

Georgia officials pushed back on widely circulated reports about Claude accessing election data, saying it retrieved only free public information. (Georgia Secretary of State)

Canadian legal experts say defendants need more resources to challenge AI evidence. (Law Commission of Ontario)

Police in India say a murder defendant used an AI-altered funeral image to obtain a death certificate and evade trial. (NDTV)

Dutch police say AI-generated bank warnings helped scammers steal €380,000 from about 170 victims in three weeks. (Dutch police)

South Korean police say an in-house AI tool helped analyze 530,000 records and identify two hacking suspects. (Seoul Economic Daily)