OpenAI is making two high-profile hires as it prepares for an initial public offering, recruiting Noam Shazeer, one of the architects of modern generative AI, and Dean Ball, a former White House AI policy official. Shazeer announced his departure from Google on Wednesday after roughly 26 years at the company, including a three-year break to co-found the AI role-playing startup Character AI. Google paid $2.7 billion in 2024 to bring him back and gain access to Character AI's technology. At Google DeepMind, he served as a co-lead on the Gemini project, cementing his reputation as a foundational figure in the field. Shazeer is one of the co-authors of the 2017 paper "Attention Is All You Need," which introduced the Transformer architecture that underpins today's most powerful AI systems. The personnel move is part of an ongoing shuffle of elite researchers among OpenAI, Google, Anthropic, and Meta, all racing to lead the next phase of AI development.
Shazeer's exit from Google wasn't entirely quiet. According to The Information, he posted opinions on internal message boards about transgender identity and Israel's war in Gaza that were inflammatory enough that management deleted the messages. Whether those controversies will resurface at OpenAI remains an open question, particularly as the company courts public market investors sensitive to governance and reputational risk.
Ball, meanwhile, brings Washington experience to OpenAI's policy shop. He served briefly on the White House team that helped produce America's AI Action Plan before stepping down to rejoin the Foundation for American Innovation, a techno-libertarian think tank, as a senior fellow. Starting July 6, he will lead a new "Strategic Futures" team reporting to Chief Strategy Officer Jason Kwon. In a blog post, Ball said the small group will tackle "catastrophic risk, recursive self-improvement, labor market impact, and the relationship between the frontier labs, governments (particularly the U.S. Federal Government), and society," handling both public-facing policy and internal governance.
Ball's framing is notable for the burden it places on AI labs themselves. He wrote that "almost by necessity" the companies building frontier models will have to take the lead on governance, suggesting internal oversight will be just as important as the rules governments ultimately set. That posture aligns with OpenAI's broader effort to position itself as a responsible steward of powerful AI systems, a narrative that will likely matter a great deal when the