Latest news

Search and community filters apply to these 50 stories in this part of the archive. Continue to earlier reporting below.

Business

Berkeley AI Institute METR Red-Teams Anthropic Agent Monitors and Reviews Claude Opus 4.6 Sabotage Report

Berkeley

METR researcher David Rein uncovers monitoring bypass vulnerabilities during a three-week red-team audit of Anthropic's infrastructure…

2 min read

METR Fine-Tuning Experiments Show Small Datasets Sharply Increase Chain-of-Thought Controllability

Researchers Kei Nishimura-Gasparian and Neev Parikh demonstrate that fine-tuning reasoning models on small instruction datasets…

2 min read

Search earlier stories