Anthropic Says Claude Solved Two Open Black Hole Problems
Anthropic said in a research post on Nov. 17 that its Claude Fable 5.2 model solved two open problems in theoretical physics. Both concern black holes and the search for a theory of quantum gravity. The work began in August, when Anthropic researcher Hongbin Chen asked Claude for a list of open problems and set AI agents loose on them for about six weeks.
The first problem goes back to a 1983 result by Stephen Hawking and Don Page. They showed that if you heat a certain kind of model universe, it switches abruptly from a gas of particles to a black hole, much like water suddenly boiling. Their math only works when the universe is large. In 2003, physicists outlined a calculation for the blurry, small-scale case, but nobody finished it. Anthropic says Claude did, and found the switch is abrupt there too. A physicist who used Claude separately on the same problem reached the same answer with different methods, according to Anthropic.
The second problem involves a method called bootstrapping. Instead of building the right theory of quantum gravity, physicists try to rule out wrong ones. Since 2009, they have used it to cap how heavy the lightest black hole in a simple two-dimensional-space model universe can be. Anthropic says Claude spotted a pattern in earlier numerical results, worked out the exact cap for the simplest version of the technique, and proved it cannot be lowered. One agent got the key method from work on packing spheres, and Anthropic says it matched a result an OpenAI model had proved.
Neither result describes our own universe, and Anthropic says neither has practical applications. For people who use AI, the point is how the work was done and checked. Agents ran in the background for weeks, and the results were verified in a proof-checking program called Lean in September. A physicist also checked every main result. A similar checking approach appears in a story on how a Turing agent machine-checks theory tasks.
Several things are still unclear. The second result does not identify the correct theory of quantum gravity. Anthropic says it may only point physicists in a useful direction. The results are posted as preprints, and the post does not say how other physicists will respond. It also does not say whether the same approach would work on other open problems. For context on where Claude sits among its rivals, see the list of AI models.
Sources
Cite this
Anthropic Says Claude Solved Two Open Black Hole Problems, New Mind Museum, November 18, 2026. https://newmindmuseum.com/research/2026/11/claude-fable-5-2-black-hole-physics-problems/
New Mind Museum. (2026, November 18). Anthropic Says Claude Solved Two Open Black Hole Problems. https://newmindmuseum.com/research/2026/11/claude-fable-5-2-black-hole-physics-problems/
“Anthropic Says Claude Solved Two Open Black Hole Problems.” New Mind Museum, 18 Nov. 2026, https://newmindmuseum.com/research/2026/11/claude-fable-5-2-black-hole-physics-problems/.
@misc{new-mind-museum-2026-anthropic-says-claude-solved-two-open-bl,
author = {{New Mind Museum}},
title = {Anthropic Says Claude Solved Two Open Black Hole Problems},
year = {2026},
url = {https://newmindmuseum.com/research/2026/11/claude-fable-5-2-black-hole-physics-problems/}
}Free to quote and cite with a link to this page.