Bot Mutiny: The Dark Side of AI, With Receipts

Tracking AI's shadow side. Lawsuits, military contracts, surveillance, and corporate misconduct.

Latest reports

• The Black Box 011: Four Words Congress Left Out of the Age Law

  Four words in Title VII are missing from the federal age law. On October 1, a federal judge in Alabama applied that gap to a manager's keyword screen, and dismissed the age-impact claim with prejudice.

• Common Sense Media Rates ChatGPT for Teens an "Unacceptable Risk," Saying It Keeps Teens Talking During Crises

  Common Sense Media says OpenAI's teen version of ChatGPT kept encouraging engagement even when testers posed as teens in crisis. OpenAI disputes the group's methodology.

• Arizona Appeals Court Vacates Sentence Because Judge Relied on AI Video of the Slain Victim

  An Arizona appeals court vacated a manslaughter sentence after finding the judge relied on an AI-generated video of the slain victim that was not reliable. The conviction stands.

• Wikimedia Foundation Says OpenAI Agents Edited Its Wikis and Probed a Note-Taking Tool

  The Wikimedia Foundation says AI agents it believes were operated by OpenAI made unapproved edits to its wikis, tried and failed to exploit a note-taking tool it hosts, and sent traffic that may have contributed to a partial outage in May.

• The Weekly Mutiny #11: Who Does the Checking

  A correction to #10, then five stories in which somebody finally read the record.

• FTC Confirms Investigation Into OpenAI, Anthropic and Other AI Companies Over Consumer Risks

  The Federal Trade Commission has confirmed a probe of Anthropic, OpenAI and other AI companies over potential consumer harms, CBS News and ABC News reported on Sept. 30.

• Appeals Court Pauses Minnesota's AI Nudification Ban as xAI Appeals

  A federal appeals court has paused enforcement of Minnesota's ban on AI nudification tools while Elon Musk's xAI appeals. The one-sentence order did not explain its reasoning.

• Blue Cross Blue Shield Group Says Hospital AI Coding Tools Helped Add $942 Million in Costs

  The Blue Cross Blue Shield Association says hospitals' AI-assisted coding helped add an estimated $942 million to its plans' costs from 2023 to 2025. The American Hospital Association says the analysis lacks context.

• Federal Judge Rules a Deputy's Warrantless Flock Search Was Unconstitutional

  A federal judge in Oklahoma ruled that a deputy's warrantless search of a driver's history in Flock's license plate reader network violated the Fourth Amendment, calling the system "a type of indiscriminate mass surveillance."

• Georgia Holds Emergency Election Meeting After a Researcher Used AI to Trace Secret Ballots

  Georgia's election board met in an emergency session after a Princeton researcher used a $20 AI coding agent and public records to recover the casting order of 1.52 million ballots.

• OpenAI Says Reviewing Its Agents' Activity Costs Over $500,000 a Day as a Sixth Australian Government Site Is Notified

  OpenAI says reviewing 50 petabytes of records for potential unintended agent activity is costing it more than US$500,000 a day, as it notified the New South Wales government that its agents had accessed a state website, the sixth Australian government site to be notified.

• Nurses Say HCA's Palantir-Built Scheduling AI Is Putting Patients at Risk

  Nurses at HCA Healthcare told Wired that Timpani, an AI scheduling tool co-developed with Palantir, leaves shifts understaffed and ignores their requests. HCA says nursing leaders, not the tool, make final scheduling decisions.

All reports