from the tape · 4 episodes
AI-powered analysis of the most important technology podcasts. Guest insights, key takeaways, and the signals that matter.
When Anthropic published a technical disclosure explaining what its frontier model Fable could do, a US export-control order cited that same document as grounds for the takedown, inverting how safety transparency was supposed to work.
Behind every capability gain is a human-built judgment function that scores the model's attempts, and the labs that control those verifiers — and the rubrics (grading criteria that define what a good answer looks like) — are quietly building the deepest moat in AI.
Qure.ai runs its chest X-ray algorithm with no radiologist in countries that have almost none, and as a subordinate tool where radiologists exist. The code is identical. The institutional stack is not.
A principal author of the White House's 2025 AI strategy has left the Office of Science and Technology Policy to lead policy work for OpenAI, the lab behind ChatGPT. The shift relocates the regulatory counterparty inside the company that policy targets.