Fields Medalist Jacob Tsimerman joins OpenAI to study AI safety

In short: Mathematician Jacob Tsimerman says higher mathematics can help make advanced AI safer, and he is moving into AI safety research at OpenAI.

What happened

Reports say Jacob Tsimerman, a recent Fields Medal winner, is shifting part of his work toward AI safety at OpenAI. AI safety means trying to make sure AI systems do what people intend and do not cause harm.

Tsimerman’s main point is that math can give stronger assurances than testing alone. Testing is like taking a car for a few drives and hoping it will always behave. A mathematical proof is more like a step by step check that a design must work under the rules you set.

He has argued that as AI gets more capable, the risk of serious harm increases. He is not saying today’s AI is already uncontrollable. He is warning that the danger could rise as systems become better at planning, persuading, and working with other systems.

Tsimerman has also pointed to the challenge of “multi-agent” AI systems, meaning several AIs working together like a team. He has suggested mathematicians could help create proofs that keep these systems from taking unintended actions. He also co-authored a 2025 paper discussing extreme risk scenarios, including cases where AI could contribute to large-scale harm.

Why it matters

As AI tools are used in more places, people will want clearer answers to a simple question, will this system reliably stay within limits. Tsimerman’s view is that math, along with other safety work, can help build those limits earlier, before more powerful AI becomes widely deployed.

Source: NYTimes