The Information's TITV Why Oracle is Cutting Thousands, Claude Code Leak, Lessons from Google Deepmind CEO Demis Hassabis
11 snips
Apr 1, 2026 
Sebastian Mallaby, author who interviewed Demis Hassabis about DeepMind; Molly Welch, Radical Ventures partner focused on AI investing and physical-world AI; Catherine Perloff, Amazon reporter covering Rufus chatbot ads. They discuss Oracle’s cuts and AI spending. They unpack Amazon’s sponsored prompts and ad tradeoffs. They review Anthropic’s Claude code leak and risks around AI-generated code and security.
AI Snips
Chapters
Books
Transcript
Episode notes
Rufus Ads Are Cheap Research Not Big Sales—Yet
- Early Rufus ad performance is modest but cheaper than Amazon's standard ads and valuable as behavioral research.
- Marketers use sponsored prompts like 'tea leaves' to learn what chatbot users actually ask about products.
Anthropic's Double Leak Exposed Claude Code Features
- Anthropic accidentally uploaded a public developer file revealing CloudCode internals and upcoming Mythos features.
- Aaron Holmes reports this was human error and followed an earlier accidental blog post leak the same week.
Leak Shows Mythos Will Add Always-On Agents
- The leak revealed 'Kairos' always-on agents, collaboration features, and Tamagotchi-like desktop avatars for Claude Mythos.
- Aaron notes Kairos would let agents act proactively in the background, increasing autonomy for coding workflows.
