Madison Mills
7 Aug 2026Demis & Nikita Step Down | Meta’s Muse Code | Felony Benchmark | Telepathy Project
Madison Mills (Axios) on the Demis Hassabis Scoop
Axios senior AI reporter Madison Mills explains that Google embargoed the Demis Hassabis news about an hour before publication, and shares what DeepMind employees told her: low morale, a feeling of losing a CEO without a replacement, unease over the US military relationship, and a sense the company is moving in a more commercial than scientific direction. She also covers the ~$200B stock drop, the decision to bundle the Jeff Dean announcement the same day, Meta's Muse trade-off of near-free usage for training data, and Google going free-cash-flow negative versus Meta's capex discipline.
6 Aug 2026The AI Talent Wars Explained by the Reporter Who Broke the Story
Madison Mills on Her AI Talent Wars Scoop
Hosts welcome Madison Mills, whose AI talent wars scoop took over X, and ask what prompted the investigation. She traces it to Thinking Machines co-founder Lillian leaving the company citing health reasons and then rejoining OpenAI per The Information. Mills says co-founder departures to competitors feel different from past tech buildouts, and a former senior Anthropic source told her Dario had been vocal in private meetings that people were joining for money rather than mission.
Money vs Mission: Job Fears, IPO Timing, Health Leave
The hosts push back that labs offering the biggest packages in Silicon Valley can't credibly complain about people joining for money. Mills says the less-viral finding is that people who moved between labs privately worry AI could take their jobs and want to secure generational wealth before an IPO. She adds that imminent Anthropic and OpenAI IPOs and wanting to be at the winner of the AI race drive loyalty, and that former executives claim some employees take health leave to extend their time at a company before shares vest.
Anthropic's Mission-First Interviews vs DeepMind's Rift
Mills says the people she spoke to who recently left or joined Anthropic are intensely mission-focused, with moral-quandary questions built into the interview process. By contrast, Google DeepMind employees she interviewed for a previous story are at odds with Demis over Google making deals with the US military. The hosts note the dynamic of lab leaders and their employees publicly disagreeing.
How Frontier Labs Balance Pay, Talent and Burnout
Mills says she doesn't envy frontier lab leaders who must pay heavily to attract top talent while also wanting mission-driven hires. Sources told her the compensation reflects grueling workloads, including one person who said they hadn't eaten in weeks at a frontier lab. She says that's why some researchers say they're pro-making-more-money.
The Group Chats Connecting Researchers Across Labs
Mills compares researchers talking across rival labs to journalists being friends in group chats, describing young technical transplants to Silicon Valley, including one guy who lived on the couch of a future Cursor co-founder. She says elite investor group chats are seen as the creme de la creme, closely guarded, and that reporters cultivate sources inside them for breaking news.
Will the Talent War End? Diminishing Model Returns and Outro
Asked whether the talent war will end, Mills cites Zach Cass, formerly in go-to-market at OpenAI, and his idea of diminishing model returns: most people won't care whether they use GPT-7 or GPT-5, which could cool competition between labs. The hosts argue open source covers everyday needs while frontier models — like one claimed to solve ten of the hardest math problems for $2,000 of API credits — will keep demand infinite, and Mills adds AI could create whole new revenue streams. The interview closes with plans to bring her back as the show's US correspondent.