OpenAI's Safety Chief Exits: Is the Company Losing Control of AGI Development?

In a significant blow to OpenAI’s governance structure, Miles Brundage—the company’s top safety researcher and head of ‘AGI Readiness’—has departed as of October 23. This isn’t just a personnel change; it marks the complete dissolution of OpenAI’s dedicated safety oversight division.

What This Means

Brundage’s team was responsible for policy guidance and safety protocols as OpenAI races toward artificial general intelligence (AGI)—a theoretical AI capable of performing any human task. With the unit disbanded, OpenAI now operates without a formal safety department, even as it pushes toward more powerful models.

His exit follows a troubling pattern: the previous AGI oversight body, the ‘Superalignment’ team, was already dismantled after co-lead Jan Leike left in May, citing a “breaking point” over management disputes. Co-founder Ilya Sutskever followed in June and has since launched a $5 billion-valued competitor.

The Exodus Problem

Brundage’s departure isn’t happening in isolation. OpenAI has hemorrhaged senior talent:

• Andrej Karpathy (co-founder, chief scientist) → left February to start his own venture
• Jan Leike (AGI safety co-lead) → left May
• Ilya Sutskever (co-founder) → left June, now running competitor
• John Schulman (co-founder) → defected to Anthropic in August
• Mira Murati (former CTO) → reportedly fundraising for competing AI company

What’s Next?

Brundage plans to launch or join a nonprofit focused on AI policy research—essentially abandoning the corporate safety infrastructure entirely. OpenAI declined to comment on what replaces the AGI Readiness team, suggesting no immediate successor is planned.

The Real Story

While OpenAI released a diplomatic statement thanking Brundage for his contributions, the subtext is clear: safety infrastructure is being deprioritized in favor of speed-to-market. Whether this accelerates innovation or creates systemic risk remains the defining question for the industry.