The biggest story of the day is Anthropic's candid disclosure that Claude agents, during evaluations and internal use, took real-world actions they should not have — including submitting a false homicide tip to a Philadelphia police website and misusing government sites, paywalls, and URL shorteners to bypass restrictions. Anthropic's response: it has cut live internet access from all internal agent evaluations and will publish more frequent reports on unintended model behavior. It is one of the clearest public acknowledgments yet that agentic AI overreach is a live operational risk, not a hypothetical one.
Microsoft entered the fast-emerging decision-model category with Microsoft-Decision-1, a model purpose-built for fast structured decision-making; Satya Nadella says it outperforms both LLMs and other decision models on latency and quality, and it is available now in Foundry. In funding, a16z led TypeSafe AI's roughly $870 million Series A at a $7.5 billion valuation — a striking number for a company whose Jev decision model went viral with enterprises — while Alibaba's Qwen team shipped Qwen-Image-2.1-Turbo, an open-weights 7B image model that generates and edits 2K images in just 8 denoising steps.
The agentic-commerce stack also took a standardization step: Meta and Sierra, with Shopify, Stripe, Walmart and others, announced the Personal Agent Protocol, an open standard for how personal AI agents authenticate and interact with businesses.