Ex-DeepMind and OpenAI Researchers Say AI Oversight, Not Slowing Down, Is Key

In September 2026, Rishub Jain, a former Google DeepMind researcher who spent seven years there, and Daniel Kokotajlo, a governance researcher who left OpenAI in 2024, said in an interview that slowing down AI development is not the right approach. Instead, they argued, the industry needs reliable oversight and evaluation mechanisms.

Jain, who left DeepMind this year to found a nonprofit focused on AI safety, monitoring and evaluation, said external AI safety research now has more talent, funding and attention, allowing him to pursue larger-scale work outside big labs. He noted that Anthropic announced in the third week of September that its Claude chatbot now "leads" more than a quarter of the company's AI research work, an example of how AI is increasingly involved in developing next-generation models.

Kokotajlo, now executive director of the AI Futures Project, shares a similar concern in his "AI 2027" white paper: as AI accelerates research, humans could become the slowest part of the loop. More than 100 AI researchers and scholars signed a public initiative on September 18 calling for independent third-party evaluation, with access comparable to senior internal safety staff and protection from suppression of unfavorable findings.