Math Genius Tackles AI Risks: New Institute Launched

Jacob Tsimerman, a Fields Medal recipient, is shifting his focus from pure mathematics to artificial intelligence safety. Tsimerman, who announced this pivot upon receiving his prestigious award, will be taking a leave from academia to join OpenAI's safety department.

He also announced the formation of the Mathematical AI Safety Institute (MAISI), an independent research institution based in the Bay Area. MAISI is scheduled to commence its first research semester in January 2027, with plans to hire a significant number of mathematicians.

Tsimerman asserts that AI safety is a meaningful area of research and a legitimate intellectual pursuit. He highlights that while engineers implement ideas, new mathematical concepts are needed to ensure AI safety, as AI systems are fundamentally built on mathematical principles.

Mathematicians see opportunities to use advanced math to address AI cooperation, verification of responsible behavior, and the design of resilient AI systems. Concepts like zero-knowledge proofs are being explored for verifying AI agent trustworthiness. This growing ecosystem of mathematicians focusing on AI risks is fostering new research agendas and calls for third-party checks and balances.

Tsimerman also emphasizes the public's right to expect high safety standards from AI. Experts believe a multifaceted approach is necessary, with mathematics potentially offering a crucial solution.

Meanwhile, the Institute for Responsible Superintelligence (RESI), another math-oriented AI safety institute, launched in Cambridge, Massachusetts, in August. Its founding team includes notable cryptographers and AI ethics researchers.