Ep. 8 — Anthropic maps the room where Claude thinks, American robots go to war in Ukraine, and Google opts you into training its AI

Anthropic maps the room where Claude thinks, American robots go to war in Ukraine, and Google opts you into training its AI 00:48 — Anthropic finds a 'global workspace' inside Claude — a private channel where the model thinks 02:17 — The first American autonomous ground vehicles are fighting in Ukraine — and mostly being teleoperated 03:47 — If you use Google, you're helping train its AI — and it quietly opted you in 05:17 — Yesterday's 'first AI-run ransomware' still needed a human at the wheel (follow-up) 05:42 — Savi raises $7M to fight AI scams — including a live agent that listens in on your call 06:06 — Midjourney's founder: the new coding models leave his friends 'extremely productive and extremely drained' 06:30 — Vercel's CEO wants to split the model from the agent — and be 'the AWS of this generation' Daily Prompt with Archer & Iris.