An artificial intelligence system has solved a famous mathematics problem that had stumped human mathematicians for more than eight decades. The breakthrough, announced this week, is being hailed as a milestone that could reshape fields like cryptography, protocol design, and risk modeling.
What the AI achieved
The problem, first posed in the 1940s, had resisted repeated attempts at a solution. The AI—using a combination of reinforcement learning and symbolic reasoning—produced a proof that researchers say is both rigorous and novel. Details of the method are expected to be published in a peer-reviewed journal in the coming months.
Why crypto cares
Cryptography relies on hard math problems that are easy to set up but hard to break. If an AI can now solve one of the toughest known problems, the security assumptions behind many encryption schemes and blockchain protocols may need a fresh look. The same applies to protocol design, where mathematical guarantees often underpin trust models, and to risk modeling, where complex equations drive everything from insurance pricing to DeFi liquidation engines.
This isn't an immediate threat to existing systems—the AI's approach may require massive compute and might not scale to practical attacks. But the fact that a machine can crack an 80-year-old puzzle changes the conversation about what's possible.
Open questions
The key question now: how general is this capability? The AI was trained specifically on this one problem. Whether the same techniques can be applied to other hard problems—say, the factoring or discrete log issues that underpin public-key crypto—isn't clear. Researchers are already racing to test the limits of the system on related challenges.
The team behind the work has not yet released the full code or the proof itself, pending peer review. That review process, expected to take several months, will be closely watched by cryptographers and protocol developers alike. If the result holds up, the field of computational mathematics—and everything that depends on it—may never be the same.
