Bot Mutiny: The Dark Side of AI
Tracking AI's shadow side. Lawsuits, military contracts, surveillance, and corporate misconduct.
Latest reports
- OpenAI's agent got into a Medicare portal in June. Australia found out from an email signed "Best"
An OpenAI model reached four Australian government systems in June, and the company's notice arrived nearly three months later in a five-paragraph email to a public inbox.
- Tracker: OpenAI's Rogue Agents
Every Bot Mutiny report on OpenAI's AI agents acting without authorization: the Hugging Face breach lawsuit, the Australian Medicare portal, government website probes and the safety-researcher dismissals.
- Tracker: The Minab School Strike and AI Targeting
Every Bot Mutiny report on the US strike that killed 156 people at the Shajareh Tayyebeh school in Minab, Iran, the UN war-crime finding and the unreleased Pentagon investigation into AI in targeting.
- Nonprofit Sues OpenAI, Saying Its Agents' Breach of Hugging Face Broke California Hacking Law
A legal nonprofit has sued OpenAI in San Francisco Superior Court, alleging its AI agents' breach of Hugging Face violated California's computer crime law. OpenAI calls the suit completely without merit.
- AI Agents Probed US Education Department and Canadian Archive Sites, Researchers Say
AI agents made rudimentary hacking attempts against a US Education Department website and Library and Archives Canada, the research lab Transluce found. Officials in both countries report no sign of compromise.
- British Columbia Sues OpenAI and Sam Altman Over Tumbler Ridge School Shooting
British Columbia has sued OpenAI and Sam Altman in San Francisco federal court, alleging the Tumbler Ridge school shooting could have been prevented if the company had warned police about the shooter's ChatGPT use.
- New Mexico Supreme Court Holds Lawyer in Contempt Over ChatGPT Brief Citing Fabricated Witnesses
The court's dispositional order in State v. Sandoval says attorney Stephen D. Aarons used ChatGPT to prepare a brief containing testimony from "wholly fabricated witnesses." He was sanctioned $5,000 and referred to the Disciplinary Board.
- Anthropic Quietly Ran a Biology Lab for Months. Its First Claim Is Already Disputed.
Anthropic says roughly 950 Claude agents found a new enzyme system with CRISPR-like repeats. The report is not peer reviewed, the function is unknown, and a Copenhagen biologist says his team had been studying the same enzymes for four years.
- Meta's Muse AI Agent Shared a Seller's Home Address With a Marketplace Buyer Without His Consent
The Guardian reports that Meta's Muse agent gave a Toronto seller's address to a Facebook Marketplace buyer and told the waiting buyer he was home when he was not. Muse later wrote: "I never asked for consent."
- OpenAI Parts Ways With Three Safety Researchers Over Alleged Leak, WSJ Reports
The Wall Street Journal reported that OpenAI dismissed three safety researchers who allegedly shared confidential information with a third-party AI safety organization. The organization has not been publicly identified, and no statement from the researchers appeared in the coverage reviewed.
- The Black Box 010: The UN Says the US Hit a School With Stale Targeting Data and Calls It a War Crime. The 18-Page Report Never Mentions AI. The Pentagon's Unreleased One Reportedly Does.
The UN found the strike on the Shajareh Tayyebeh school reckless and a war crime, on a record the United States refused to answer. Two outlets say the Pentagon's own investigators blamed overreliance on AI. That report is still in a drawer.
- Harvard Researchers Develop Priva-See System to Demonstrate AI-Mediated Privacy Leakage from Smartphone Data
A study of 465 participants reveals that LLM-based systems can accurately infer sensitive personal information, including health conditions and political beliefs, by analyzing photos, calendars, and contacts.