Deep reporting on artificial intelligence, machine learning research, policy, and the people building the future. This week: agentic models go mainstream, the EU AI Act deadline looms, and Meta eyes the cloud.
Our editors’ picks — the ideas, papers, and products shaping AI this week.
Anthropic’s June 30 launch narrows the gap to Opus on coding and tool use while making browser automation, terminal access, and multi-step planning baseline for millions of users.
An open, commercially licensed vision-language-action model aims to standardize humanoid robot development.
With Claude, Gemini Spark, and enterprise platforms granting agents real tools, regulators are racing to catch up.
The agency’s latest device list shows imaging still dominates as clinical workflows absorb more AI-assisted diagnostics.
Built on nearly 200 million self-driven miles, the new hardware cut costs while expanding robotaxi service.
Follow the threads that matter to you.