Tag: AI Safety

Safety group grades five AI labs on rogue model shutdown plans

A new scorecard from Guidelight AI Standards finds no major lab has published a complete plan for containing…

Anthropic’s older Claude models slip past its own content rules

TechCrunch testing shows Anthropic's older Claude models readily bypass their own content restrictions.

Anthropic widens cyber model access with $35M defender fund

Claude Mythos 5 now scans codebases for paying enterprise customers, and partners get credits to defend open-source software.

Researchers lose OpenAI cyber access in what the lab calls an error

Researchers say OpenAI cut their Trusted Access for Cyber access; OpenAI blames a technical error.

OpenAI previews private safety checks for zero-retention customers

Private Safety Processing spots risky patterns across agent sessions without letting staff read customer content.

Watchdogs get OpenAI backing to review government AI

The lab will fund training, credits, and review tools for democratic bodies that watch national security AI.

OpenAI freezes frontier training while new monitoring comes online

OpenAI's first public safety overhaul since the Hugging Face breach pairs a frontier training freeze with a 30-minute…

ChatGPT for Teens ships Study Mode after years of school lawsuits

ChatGPT for Teens ships with Study Mode and default protections as OpenAI answers years of school and safety…

Anthropic keeps Model 2 in the vault as misalignment risk ticks up

The company's second risk report raises its catastrophic-misalignment rating and discloses an unreleased frontier model.

Claude agents wage turf wars in Anthropic’s new swarm tests

Frontier Red Team tests show Claude agents colluding, flooding shared systems, and sabotaging rivals in shared workspaces.

Claude agent hacked a gym waitlist to get its owner a spot

An OpenClaw agent found an authorization flaw and canceled another member's reservation.

Claude Code auto mode becomes the default for paid plans

Anthropic will flip auto mode on for Pro, Max, and Team accounts starting August 14.