Just weeks after receiving the Fields Medal—mathematics' highest honor, often dubbed the subject's Nobel Prize—Jacob Tsimerman is stepping away from the discipline entirely. Rather than using the prestigious award to solidify his trajectory as a world-leading mathematician, the University of Toronto professor is joining OpenAI to work on artificial intelligence safety, driven by concerns over the catastrophic risks posed by rapidly advancing AI systems.
Tsimerman's decision comes at a moment when mathematicians are increasingly astonished by the speed at which AI models are resolving long-standing open problems. While some in the field worry about their profession becoming obsolete, Tsimerman believes the existential stakes for humanity are far higher. Having personally witnessed AI resolve complex mathematical conjectures he once considered tackling, alongside recurring real-world security breaches, he concluded that urgent intervention is required.
In 2025, Tsimerman co-authored a paper examining scenarios where unconstrained AI could lead to human extinction. The paper outlined concrete pathways to catastrophic outcomes, ranging from unintentional devastation—such as an indifferent AI deprioritizing global food agriculture—to intentional civilizational conflict and escalations triggered by constrained models breaking their online bounds. By delineating these specific, tangible failure modes, Tsimerman aimed to move the safety debate away from speculative philosophy and toward actionable, rigorous counter-measures.
Known for his ability to uncover surprising bridges between disparate mathematical fields—such as applying "o-minimality" to make major progress on the Hodge conjecture—Tsimerman hopes to bring fresh analytical rigor to AI alignment. He argues that current evaluation protocols, which rely heavily on asking models to act benignly and observing their behavior over short windows, are insufficient. Instead, he advocates for standardized, non-bypassable verification constraints and continuous testing regimes administered by independent third parties alongside international regulatory bodies. Although acknowledging a sense of personal grief in leaving a discipline deeply tied to his identity, Tsimerman emphasizes that protecting human safety must take precedence.