Anthropic reports Claude writes more than 80 percent of its merged code and warns of recursive self-improvement risks that could outpace human oversight.
// daily-digest · sat 06 jun 2026 · morning refresh · last 72h focus
Seven heavy-lane items on AI self-improvement, strategy, fintech, and capital markets, one rationalism piece, one Ruby roundup, and one archaeology wildcard.
— refreshed for you, in 10 items.
Anthropic reports Claude writes more than 80 percent of its merged code and warns of recursive self-improvement risks that could outpace human oversight.
Thompson compares Nvidia’s AI PC positioning with Microsoft’s Build vision for agentic devices and where computing leverage is shifting.
Interactive map of 170-plus companies building autonomous purchasing and agentic commerce technology, filterable by funding stage and commercial maturity.
Morgan Stanley enables external AI agents to directly access its ShareWorks and Equity Edge stock compensation platforms, among the first Wall Street firms to do so.
Google DeepMind releases Gemma 4 12B, an open-source multimodal model optimized for consumer laptops and desktops with 16GB or more VRAM.
Weekly recap of the largest venture rounds, led by Ramp’s $750 million raise alongside three $500 million rounds in AI, space, and longevity.
Willison covers OpenAI’s new Lockdown Mode feature designed to prevent data exfiltration from prompt injection attacks in enterprise deployments.
Scott Alexander opens reader voting to narrow 162 submitted book reviews down to a dozen finalists for the 2026 annual contest.
Weekly roundup covering Bundler 4.0.13’s gem security cooldown feature, Ruby AOT compiler improvements, and SimpleCov 1.0 RC1 release.
Scientists cultured ancient yeasts from Ötzi the Iceman’s body to recreate a functional 5,000-year-old sourdough starter.