A Famous Developer Merged Eight PRs Without Reading Them — Because the AI Approved. The Accountability Vacuum Is Now Real.
Theo says he merged eight pull requests on the strength of an AI reviewer's approval alone, without reading the code. Developers are now openly debating who is responsible when the gate is a bot.
The most quietly alarming story of the week isn't a model breakout or a superintelligence prediction. It's a workflow. As @connect24h reported, the well-known developer Theo posted that he merged eight pull requests on the basis of Codex-driven "Ox Alpha" approval alone — without reading the code himself. The framing in that thread cuts to the heart of it: AI review is no longer an assistant that flags issues for a human. It has begun to *become* the PR gate itself.
This is a categorical shift, not an incremental one. For the last two years, the industry consensus on AI code review was that it augmented human judgment — a second set of eyes that caught what tired engineers missed. What Theo describes is inversion: the human is no longer in the loop at all, and the AI's approval is treated as sufficient. When one prominent developer does this, it's a personal risk tolerance. When it's posted approvingly and spreads, it's a norm forming in real time.
Get our free daily newsletter
Get this article free — plus the lead story every day — delivered to your inbox.
Want every article and the full archive? Upgrade anytime.
No spam. Unsubscribe anytime.