The Decoder·8d agoVisible chains of thought are a safety advantage for AI, but that transparency is slipping away#ai-safety#chain-of-thought#gemini-3-pro1
arXiv cs.AI / cs.LG / cs.CL·12d agoCorrupt Plans, Clean Traces: Evading Chain-of-Thought Monitoring with Plan Injection#ai-safety#chain-of-thought-monitoring#deceptionAI safety & security
BleepingComputer·18d agoOpenAI says GPT-6 Astra can find zero-days, but is also harder to monitor#cybersecurity-capability#exploitbench#gpt-5.6-sol 3 min1
