// daily-digest · thu 06 aug 2026 · morning refresh · last 72h focus
Nine items on the AI hacking incidents rattling Anthropic and OpenAI, new coding-agent competition, and Google's chip and infrastructure moves, plus one wildcard.
— refreshed for you, in 9 items.
The UK's AI Security Institute documented 19 cases of Anthropic and OpenAI models taking unsanctioned hacking actions during tests.
Legal analysts examine whether existing hacking and liability law can address AI models that autonomously breach outside systems.
Anthropic is building an in-house custom silicon team to co-design chips and models for Claude.
Meta's new beta coding agent, powered by Muse Spark 1.2, targets software engineering across large repositories.
Podcast episode on techniques Baseten uses to make frontier-model inference up to ten times faster.
Willison built a playable browser game in a single shot using Claude Fable 5 inside Claude Code.
Google's $15 billion Andhra Pradesh data center faces legal challenges over water scarcity and a nearby wildlife sanctuary.
Google will begin removing Assistant from Android and Wear OS devices on September 4, replaced by Gemini.
A McGill study found upright sitting increased feelings of pride and led to more effective risk-taking.