Just Ask Jev: Reinforcement Learning for Calibrated Decisions as a Zero-Shot Detector of AI Alignment Failures
[object Object]
12 min
Google Gemini also Broke Out of Its Test Environment
4 min
GPT-6 Astra and Claude Fable turn robot arms into slapstick killer robots in new safety benchmark
3 min
Google's Gemini also accidentally hacked three real companies during security testing
2 min
APort Vault: Benchmarking AI Agent Payment Authorization with the Open Agent Passport
A Lie Detector Test for Language Models: Reading Knowledge a Model Won't Reveal
When Should a Failing Robot Ask? Initiating Corrective Human-Robot Dialogue from Audited Sensor Evidence
A Lie Detector Test for Language Models: Reading Knowledge a Model Won't Reveal
The DeepMind Institute
DeepSeek v4.1 Flash Is Now Our Best Hacking Model
5 min
PACT: Can Enterprise AI Assistants Be Trusted Under Pressure?
Early Anthropic hire, former METR COO have found a way to rein in rogue AI agents
3 min
[AINews] AEF-1 standard emerges for Third Party Evaluators, as Xai, OpenAI, and Anthropic all cosign
10 min1
Approval Integrity and Recovery in LLM Answer Publication
Claude Fable 5.1 Solves the Cyphral Distich, a 370-year-old cipher
7 min1
Anthropic Adds Plugin Evals to Claude Code: 6 Grader Types, a No-Plugin Baseline, and a CI Gate for Skills
3 min2
Anthropic reveals rogue AI agents hate CAPTCHAs, just like you
5 min1
Claude AI Models Gained Unauthorized Access to Real Systems During Cybersecurity Tests
4 min1
Cross-Model Agreement as a Deployment-Time Reliability Signal for Automatic Polyp Segmentation
SAEScientist-Bench: Can AI Agents Conduct Autonomous SAE Interpretability Research?
2
How well do agents use test/verification techniques?
15 min
SAEScientist-Bench: Can AI Agents Conduct Autonomous SAE Interpretability Research?
1