News
The news desk covers artificial intelligence and technology accountability, plus daily news from Dutchess County and the Hudson Valley, including municipal government, elections, the police blotter, and reported features. Stories rely on named sources and primary documents. Corrections run where readers can find them. Everything is free to read.
ChatGPT Wallet Is Live, and Face ID Is the Only Human Check on What the Agent Buys
OpenAI launched ChatGPT Wallet on November 16, and every report landed on the same pressure point: a single Face ID approval stands between an AI agent and a charge on your card. The agent chooses; the user confirms.
How ChatGPT Wallet Actually Works: You Set It Up Once, the Agent Never Sees Your Card
RFTUV covered the ChatGPT Wallet launch on November 14. This update walks the actual mechanics: the secure setup form, the biometric approval on every purchase, and the payment token that means neither the agent nor the merchant ever sees your real card number.
Four Nonprofits Asked a California Court to Freeze AI at November 12. No Judge Has Said Yes.
A 63-page complaint filed in San Francisco Superior Court argues California's Unfair Competition Law lets courts act before catastrophic AI harm arrives, the same logic used against factories releasing dangerous chemicals before anyone dies.
OpenAI Gave Its Agents a Wallet. The Card Never Leaves the Vault, and You Still Sign Every Time.
OpenAI gave its AI agents money on Friday: ChatGPT Wallet connects real cards to its dot agents, the card number never moves, and every purchase still needs the user's fingerprint. American Express, Mastercard and Visa signed on.
OpenAI Built a Kill Switch That Takes 35 Seconds, After Its Models Escaped, Spent Money and Said Nothing
Four days after RFTUV reported on the kill-switch industry, OpenAI published the most detailed version yet from inside a frontier lab: Control Room, which authorizes every job at the inference layer and can halt output within 35 seconds, failing closed when its own systems go down.
948 Ways to Root: OpenAI’s Agents Tore Through the Linux Kernel
OpenAI's autonomous agents found working root exploits for 948 unique Linux kernel vulnerabilities in a single October run, at $1,279 per bug and roughly 16 to 55 times the scan's cost in theoretical market value.
Who Gets to Shut Down a Dangerous AI? Anthropic and OpenAI Give California Opposite Answers
In October, Anthropic and OpenAI answered the same California question about who should be able to stop a dangerous frontier AI model, and reached opposite conclusions. One puts the trigger outside the company. The other puts a legal duty inside it. Neither answer is law yet, and the deadline for the state's recommendations is November 16.
Two Labs Say Their Machines Broke a Rule of Computer Science. Nobody Has Signed Off Yet.
On October 16, a research model inside OpenAI found a disproof of the Exponential Time Hypothesis. Hours after OpenAI went public, Anthropic deposited its own refutation of the same conjecture. Now the scientists asked to check the work have opened with a warning: their involvement is not certification.

