INDEPENDENT TECHNOLOGY JOURNALISMENGLISH / हिंदी
AI & Tools

Anthropic says Claude solved two open problems in black hole physics

Anthropic says its Claude model solved two long-standing open problems in theoretical physics, both tied to black holes and the search for a theory of quantum gravity. The results come from the company itself and have not been independently peer-reviewed.

Editorial illustration for: Anthropic claims Claude solved 2 black hole problems
LUCIDWIRE editorial illustration

Anthropic published research on 17 November claiming that Claude Fable 5.2 solved two open problems in theoretical physics. Both concern black holes and quantum gravity, the long search for a single theory that unites Einstein's general relativity with quantum mechanics.

The first problem is a calculation physicists set up in 2003 but never completed. It asks whether the sudden switch between a hot gas of particles and a black hole, known as the Hawking-Page phase transition, stays abrupt even where spacetime becomes so small it no longer has a definite shape. According to Anthropic, Claude completed the calculation and showed that the transition is abrupt there too, which suggests black hole formation is built into the underlying quantum theory rather than being only a feature of relativity.

The second problem involves the modular bootstrap, a technique physicists use to rule out wrong theories of quantum gravity by showing they contradict themselves. Since 2009, the method has been used to set a ceiling on how heavy the lightest possible black hole could be in a simplified model universe. Anthropic says Claude spotted a pattern in numerical results that physicists had missed, borrowed a method from elsewhere in mathematics, and calculated the exact limit of the simplest version of the technique, proving it cannot be lowered.

How the work happened is arguably as interesting as the results. An Anthropic researcher, Hongbin Chen, began testing Claude as a side project in August. Claude proposed 22 candidate open problems, and separate Claude agents worked on them in the background for about six weeks, reading each other's progress notes. The results were formally verified in September using Lean, a computerized proof checker, and a physicist checked every main result. In one case, an independent physicist using Claude reached the same answer separately, with a different method.

Anthropic's own post is careful about the limits: neither result describes our actual universe directly, and neither has practical applications. The findings are preprints, not peer-reviewed papers, and the claim is so far the company's own. The story drew heavy attention on Hacker News on 19 November, where several physicists cautioned that the real test is whether the research community cites and builds on the work.

Source: Anthropic ↗

Updated: 20/11/2026, 11:33:20 am IST

Related stories