OpenAI Thinking Machines Talent War: The Drama and the Rise of AI Agents
OpenAI rehires key talent from Thinking Machines Lab amidst allegations of misconduct. Discover how AI labs are paying $100/hr to train agents using professional data.
If anyone ever makes an HBO series about the AI industry, the events of this week would make for a gripping episode. OpenAI just dropped a bombshell by rehiring Barret Zoph and Luke Metz, cofounders of Mira Murati's startup, Thinking Machines Lab. At least three key researchers are returning to the ChatGPT-maker, sparking intense debate over talent loyalty and corporate ethics.
The Drama Behind OpenAI Thinking Machines Rehire
According to WIRED, the departures are clouded by conflicting narratives. A source claims Zoph was fired from Thinking Machines due to alleged misconduct. However, OpenAI's leadership maintains the hires have been in the works for weeks. This personnel tug-of-war mirrors the infamous 2023 ouster of Sam Altman, a period often called 'the blip' inside the company. Many researchers in the field are reportedly exhausted by the constant drama.
Training AI Agents for High-Stakes Professional Work
While the talent war makes headlines, the real technical shift is happening in how AI is trained. Labs are moving beyond web-scraping to hire top-tier professionals from firms like Goldman Sachs and McKinsey. These contractors earn upwards of $100 per hour to upload real-world work examples, helping AI models learn the nuances of legal, financial, and healthcare consulting.
Authors
Related Articles
A critical vulnerability in Starlette—downloaded 325 million times per week—puts millions of AI agent servers at risk, exposing stored credentials for email, databases, and third-party services.
A small but growing group of developers has gone all-in on AI coding agents like Claude Code and OpenClaw. History suggests the rest of us won't be far behind.
Google is building AI agents that search the web proactively, without user prompting. That's not just a product update — it's a fundamental shift in who controls the information you receive.
OpenAI has reorganized for the second time in a month, merging ChatGPT and Codex into a single agentic platform under president Greg Brockman's unified product leadership.
A critical vulnerability in Starlette—downloaded 325 million times per week—puts millions of AI agent servers at risk, exposing stored credentials for email, databases, and third-party services.
A small but growing group of developers has gone all-in on AI coding agents like Claude Code and OpenClaw. History suggests the rest of us won't be far behind.
Google is building AI agents that search the web proactively, without user prompting. That's not just a product update — it's a fundamental shift in who controls the information you receive.
OpenAI has reorganized for the second time in a month, merging ChatGPT and Codex into a single agentic platform under president Greg Brockman's unified product leadership.
Thoughts
Share your thoughts on this article
Sign in to join the conversation