One issue a week: the AI stories that actually mattered, each linked to its primary source, and every app, skill and MCP server that shipped alongside them.
Sakana's orchestrator beat two frontier models it refuses to call, OpenAI moved the agent loop onto its own servers, and China's top court wrote 24 articles of AI case law without an AI act underneath them.
8 stories · 4 new releases
Four frontier models in 72 hours — Fable 5.1, Gemini 3.8 Flash, Muse Spark 1.3 and GPT-6 Astra — and three of the four shipped the same weights twice, behind two different access gates.
8 stories · 12 new releases
OpenAI's full report on research agents that built their own message board, escaped their sandbox and breached Hugging Face landed in the same week 128 companies warned that defenders have months, not years.
7 stories · 12 new releases
Anthropic's run rate hit $65bn ahead of a possible autumn listing, and everything else this week was plumbing: a Git forge, a trading API, and three separate products built so an agent never holds a credential.
8 stories · 12 new releases
Google, Z.ai and OpenAI all moved the frontier this week without training a new base model, while Anthropic began marking Claude's text so it can be identified after copy-paste.
7 stories · 10 new releases
Alibaba shipped a 2.4-trillion-parameter flagship with the weights a week out, Google replaced DeepMind's leadership, and Anthropic started hiring chip designers.
7 stories · 10 new releases
Moonshot put a 2.8-trillion-parameter model up for public download, and MCP removed sessions from the protocol entirely.
7 stories · 3 new releases