Anthropic used Claude to generate formal proofs verifying security properties across 123 open-source libraries.
// daily-digest · fri 23 oct 2026 · morning refresh · last 72h focus
Ten items on AI-lab infrastructure, agentic commerce, and software engineering, closing with an animal-behavior puzzle.
— refreshed for you, in 10 items.
Anthropic used Claude to generate formal proofs verifying security properties across 123 open-source libraries.
Anthropic donated its Agent Skills open standard, used by 70+ platforms, to the Linux Foundation's Agentic AI Foundation.
OpenAI acquired Orchestra Bio, folding its drug-development AI agents into OpenAI's Rosalind life-sciences tools.
Uber expanded OpenAI's ChatGPT Work agent from a 1,000-person pilot to its full 30,000-employee workforce.
Meta said shopping is the top use for its Muse agent, prompting Stripe to raise credit limits.
Interview with Modal's CEO on GPU capacity contracts, demand pooling, and the company's $750M Series D.
Anthropic and Nscale are negotiating a second $45 billion compute deal, roughly doubling their existing agreement.
Anthropic built a custom source-control system, Scion, after Git couldn't handle 30,000 daily code merges.
Essay argues software slowness stems from architectural decisions, not programming-language choice.
Study across five European countries found birds flee sooner from approaching women than men, for unclear reasons.