.@thsottiaux and @ajambrosino on why verifying code is the new bottleneck. AI is moving fast. The past week alone saw two major model drops: @OpenAI's GPT-6 Astra and Anthropic's dual release of Claude Fable 5.1 and Mythos 5.1. But faster models mean even more code and more features to verify—and tasks like de-bugging and clicking through to check that buttons work still takes real human time. Sottiaux and Ambrosino, from OpenAI's Codex team, flagged this exact challenge six months ago on…