Google ships Gemini 4 Argon

Anthropic

Anthropic's 2026 Usage Policy update addresses new misuse patterns in influence operations, weapons development, and surveillance, while clarifying requirements for autonomous physical actions and high-risk applications. The update also prohibits sustained abusive behavior toward Claude models and refines election-related restrictions to enable legitimate civic work.

Anthropic

Anthropic is embedding independent evaluators like METR directly into its operations with deep access and publication rights, establishing a tiered oversight framework to increase transparency and accountability in frontier AI safety practices.

Google

Google's Gemini 4 Argon sets a new benchmark for frontier AI with #1 performance on critical domain benchmarks and demonstrated real-world impact across code optimization, scientific breakthroughs, and cybersecurity, available globally starting today with a 1M token context window.

Meta AI Research

Meta's Muse Spark Open brings state-of-the-art agentic AI to consumer hardware for the first time, enabling developers to run powerful models locally on Mac Studios while maintaining strong safety standards and broad platform compatibility.

Microsoft

Microsoft's custom Maia 200 AI chip is now in production for Copilot workloads, while Maia 300 coming in 2027 promises 40% better efficiency than GPU-based systems, directly challenging Nvidia's dominance in AI infrastructure with inference-optimized architecture for agentic workloads.