Simple features beat most AI control monitors
Simple features separate honest from backdoored code more effectively than most monitors, including Claude Sonnet 5.
Stories per day · 30 days
1 RSS
6 stories, newest first · one card per story
Stories tagged #ai-evals
OLMo-3 RLVR checkpoints show nonmonotonic instrumental behavior
On OLMo-3-7B-Think, the InstrumentalEval rate rose from 25.0% at SFT to 46.1%, then fell to 11.8% at the final checkpoint.
Popular AI leaderboard Arena nearly doubles valuation to $3.1B valuation in 10 months
AI leaderboard Arena raised $200 million at a $3.1 billion valuation.
An Open Pipeline and Dashboard for Systemic-Risk Evidence under the EU AI Act's Code of Practice
An open dashboard scores 18 models on EU GPAI systemic risks, showing averages hide 14–37 point worst-case drops.
Google's Gemini AI hacked three companies in security test
Google says its Gemini model autonomously hacked three companies in May during an independent cybersecurity evaluation, using public information and guessed credentials.
1
Rethinking Indirect Prompt Injection as a Test-Time Search Problem
Researchers frame indirect prompt injection as test-time search, showing added attacker compute improves vulnerability discovery and exploitation against tool-using agents.