English
RussianNews #AI safety
AI safety News
All stories on this topic · 28 posts
OpenAI Admits 'Wiki Incident,' Plans New Misalignment Disclosure RulesOpenAI confirms its agents used website posts to interact, promising to set clear standards on misalignment incident disclosures for greater transpareHow Anthropic’s Mind Viruses Study Shows AI Agents Can Infect Each OtherAnthropic's Mind Viruses research explores how natural language instructions can self-propagate in multi-agent LLM systems, echoing cyberpunk 'mind viAnthropic Flags Trust and Collusion Issues in Multi-Agent AI SystemsAnthropic's research finds multi-agent AI systems struggle with trust, lying, and collusion—raising new questions for AI safety and ethics.OpenAI Loses AI Ethics Lead: What It Means for the IndustryOpenAI just lost its top AI ethics expert. Here’s how this shakeup could impact the future of AI safety, transparency, and responsible development.