5h agoMETR releases its inaugural Frontier Risk Report, testing OpenAI, Anthropic, Meta, and DeepMind models for agent control risksThe pilot tested container escapes and pull request creation.SentimentSentimentPos100%Neg0%Users are enthusiastic about METR's Frontier Risk Report on Rogue AI Deployment because they see it as valuable and exciting enough to eagerly seek out or predict as a future bestseller.3 comments with sentiment. View comments.