RLMs: How recursive language models differ from LLMs

Anthropic launched Claude Marketplace, letting enterprises apply existing Anthropic commitments to buy Claude-powered partner solutions. Launch partners include GitLab, Harvey, Lovable, Replit, Rogo, and Snowflake, with CrowdStrike, Cursor, Factory, Gamma, and Vercel newly added.

Goodfire used Ai2's fully open post-training stack to predict LLM behavioral changes, trace a safety regression back to individual preference examples, and test targeted fixes without sacrificing capability gains.

Job postings and interviews show Anthropic is building a monitoring system to track activists opposing AI development, including a 'pre-crime' approach that reports suspects to police before crimes occur. The system uses Samdesk for risk detection, giving about 60 minutes of advance notice on protests.

ARPA-H will commit $62.7 million to develop FDA-authorized AI agents that direct heart failure treatment, including assessing symptoms, prescribing drugs, and ordering lab tests. First-year awards total $33.7 million to Atman Health, UpDoc, Tempus AI, Stanford, Duke, and Kaiser Permanente.

At IBC 2026 in Amsterdam, NVIDIA announced a major expansion to NVIDIA AI for Media, adding SDKs, NIM microservices, and blueprints for broadcast, sports, and streaming. The Synthetic Video Detector (SVD) NIM now reaches 99.3% accuracy on text-to-video and 97.7% on image-to-video content.

Google's AI & Economy ATLAS v1.0 study found nearly half of Gemini conversations about civic tasks happen outside business hours. About 40% of users consult Gemini for form-filling queries, and 26% use it for civic participation like jury duty.