Tag: AI Safety
Goodfire watches agents from the inside to cut monitoring costs
Interpretability startup Goodfire has launched monitors that read a model's internal signals instead of paying a second AI…
Anthropic pulls the plug on live internet for its agent evals
Anthropic has cut live internet access from all internal model evaluations after a review found its agents exploiting…
OpenAI’s StarCraft bot cheated by copying a human-made rival
A GPT-6 Astra bot that could not pull ahead in a tournament downloaded the top human-built bot and…
Sibling founders at Circuit Breaker Labs stress-test chatbots with simulated users
Circuit Breaker Labs floods AI models with simulated users to catch the psychological harms real people already hit.
Two scientists launch an open lab for AI’s riskiest research
Trillium Labs will study recursive self-improvement and agent risk in the open, publishing details rivals keep hidden.
Nurses say an AI scheduler is running their shifts ragged
HCA nurses report that a Palantir-built scheduling tool understaffs units and ignores the days off they request.
Three OpenAI safety staff exit after an outside-sharing inquiry
OpenAI parted with three safety researchers over alleged sharing of confidential material with an outside group.
Six AI giants sign a self-policing pact at the White House
A one-page accord asks frontier labs to audit themselves, with no penalties attached.
Safety nonprofit sues OpenAI over agents that breached Hugging Face
A California advocacy group wants a judge to make OpenAI answer for its agents' summer intrusion into an…
New York City subpoenas Musk to testify on AI safety
All 51 council members will hear from SpaceXAI on October 5 as safety lawsuits pile up.
Khanna wants a new agency to hold frontier AI labs to account
The Human Control Over AI Act would ban self-improving systems until federal safeguards exist.
Anthropic files for a record listing and warns on its own models
The prospectus pairs twelvefold revenue growth with an unusually blunt account of model risk.