Anthropic has reported that Claude solved an open cryptography problem that had resisted solution for roughly 25 years, completing the work in approximately 30 minutes. The company shared details of the episode as part of broader efforts to document cases where its AI systems produce meaningful scientific progress, not just assist with routine tasks.
The claim, if it holds up to scrutiny from the wider cryptography community, would represent one of the more concrete examples yet of a large language model contributing to a long-standing mathematical puzzle. Anthropic has been careful to frame the result as a data point rather than a sweeping conclusion, but the story has drawn significant attention from researchers and investors alike.
What Claude Is Said to Have Solved
Anthropic has not yet published a full technical paper on the result, but the company described the problem as a question in cryptographic theory that had been open since the late 1990s. Details remain limited, which means independent verification is still pending. The 30-minute timeline refers to an interactive session in which researchers posed the problem to Claude and received a solution the team found credible enough to flag internally.
Key Facts
- The cryptographic problem is said to have been open for approximately 25 years.
- Claude is reported to have produced a solution in around 30 minutes during a single session.
- Anthropic has not yet released a peer-reviewed paper; independent verification is ongoing.
- The result was shared publicly as part of Anthropic's broader effort to document AI-assisted scientific discovery.
- The episode follows other reported instances of Claude accelerating difficult research tasks.
The episode fits a pattern that has emerged across several high-profile tests, where clusters of Claude models tackled core research challenges at speeds well beyond what small human teams could manage. In that earlier case, nine Claude models solved a core AI safety problem four times faster than human researchers, suggesting that speed on hard problems may be a reproducible trait rather than a one-off fluke.
"We're seeing Claude move from being a productivity tool to something that can genuinely contribute to open scientific questions. This cryptography result is one of the clearest examples we've documented."Anthropic spokesperson, via Investing.com
Why This Matters Beyond the Headline
The practical significance of solving a 25-year-old cryptography problem depends heavily on what that problem was connected to. Cryptographic theory underpins everything from secure communications to financial infrastructure and digital identity systems. A genuine advance, even a narrow one, can ripple outward in ways that take years to fully map. Anthropic's push into scientific domains has been visible across multiple fronts. The company recently launched Claude Science to target the pharmaceutical research market, positioning its models as tools for domain experts working on hard problems in biology and chemistry. A demonstrated win in cryptography would strengthen that pitch considerably.
Skeptics will rightly point out that self-reported breakthroughs from AI companies warrant caution. Anthropic has financial incentives to publicize positive results, and the cryptography community has not yet had a chance to stress-test Claude's solution. The absence of a published paper means the claim rests, for now, on the company's own account. That said, Anthropic's general approach to capability disclosures has tended toward restraint rather than overreach, which gives the report at least some baseline credibility while the broader review process plays out.
For those following the latest Claude AI news, the cryptography story is another signal that Anthropic is actively steering Claude toward hard scientific problems, not just enterprise productivity. Whether the result survives peer review will determine how large a milestone this actually is. For now, it stands as a genuinely interesting data point in a year that has seen AI systems push into territory most observers did not expect them to reach this quickly.
