Claude Solves Two Long-Standing Black Hole Physics Problems, Anthropic Says

Anthropic says its Claude Fable 5.2 model has solved two open problems in theoretical physics about black holes and quantum gravity: finishing a calculation first set out in 2003 on the Hawking-Page phase transition, and finding the exact limit of the modular bootstrap technique used to test quantum-gravity theories. Anthropic stresses that neither result describes our own universe directly or has practical uses — but both push forward a decades-long effort to reconcile quantum mechanics with relativity.

Nov 21, 2026 - 22:19
 0  6
Claude Solves Two Long-Standing Black Hole Physics Problems, Anthropic Says
A digital illustration of a black hole with a glowing orange accretion disk, its light bent by gravity over a warped blue grid representing curved spacetime, against a starry sky with the Milky Way.

Claude Solves Two Long-Standing Black Hole Physics Problems, Anthropic Says

Anthropic has reported that its Claude Fable 5.2 model solved two open problems in theoretical physics, both concerning black holes and the search for a theory of quantum gravity. The company described the results in a research post published on 17 November 2026, with preprints of the two papers posted earlier in the month.

The first result completes a calculation that physicists set out in 2003 but never finished. The second finds the exact limit of the simplest version of a technique used to test candidate theories of quantum gravity. Anthropic is careful about what the results do not mean: neither describes our own universe directly, and neither has practical applications. Their value lies in a long-running theoretical effort to understand how quantum mechanics and general relativity fit together.

What makes the announcement notable is not only the physics. Both results were produced largely autonomously by groups of Claude agents working for about six weeks, with a physicist checking every main result and the proofs formally verified in a computerised proof checker. It is the latest sign that AI systems are moving from answering questions about science to doing original research in it — a direction Anthropic has already explored, including earlier work in which Claude helped design quantum circuits with researchers at Harvard and QuEra.

A gas that becomes a black hole — suddenly, or gradually?

The first problem concerns what happens when you heat up a theoretical universe. At low temperatures, the universe fills up with a gas of particles. Above a certain temperature, a black hole forms instead. In 1983, Stephen Hawking and Don Page showed that this switch is abrupt: it is a phase transition, similar to how water suddenly boils at 100 degrees Celsius.

But Hawking and Page's calculation relies on general relativity, which holds only when the universe is vast compared with the tiny scale at which that description breaks down. As the theoretical universe shrinks toward that smaller scale, spacetime loses any definite shape and becomes blurry, and nobody knows how to do the calculation on the gravity side of the picture.

Through holographic duality, the same universe can also be described in quantum terms, with no gravity at all, written on the universe's outer boundary. On that quantum side, the calculation is possible in principle. In 1999, physicists found that a phase transition exists in that regime as well — but they did not know whether it was abrupt, like Hawking and Page's, or smooth.

In 2003, a group of physicists led by Ofer Aharony set out the calculation that would settle the question. It was extremely complex and was never completed. Whether the transition was abrupt or gradual hinged on a single number that appears only three loops into the calculation, and computing it involves many thousands of terms — far too many to work through by hand. In 2005, the same group completed the calculation for a simpler cousin of the theory and found an abrupt transition. But the version for the theory in question, N=4 super Yang-Mills theory, more than ten times larger, was never finished.

Claude finished it. Anthropic says the model ran a small research group of copies of itself, in which almost every piece of the calculation was worked out independently and checked against known results. The answer: the phase transition is abrupt. The same sudden switch appears at both extremes — the smooth, large universe of general relativity and the small, blurry one of the quantum description — suggesting that black hole formation is not just a feature of relativity but is built into the underlying quantum theory.

The result was independently cross-checked. Weeks after one of the agents reached a solution, a physicist who had independently used Claude to tackle the same problem shared his result, and a colleague realised the agent had reached the same answer earlier. The two solutions used different methods.

How heavy can the lightest black hole be?

The second problem concerns a simplified theoretical universe with two dimensions of space instead of three. Even in this simpler universe, physicists do not know how to create a full theory of quantum gravity, so they approach it from the other direction: rather than building a single correct theory, they try to rule out wrong ones by showing that they are internally inconsistent. This method is called bootstrapping.

Since 2009, physicists have used a version of this method, the modular bootstrap, to set a ceiling on how heavy the lightest black hole in such a universe could be. Any theory that breaks the ceiling is ruled out, and each lowering of the ceiling narrows the search. Physicists write the ceiling as M1 no larger than c divided by a number called Lambda, where c roughly counts how many microscopic building blocks the universe is made of. The original ceiling was c/6. Later work pushed it analytically to c/8.5, while numerical calculations hinted it could go as low as about c/9.1 — but the numbers could not prove it.

Claude noticed a pattern in the numerical results that physicists had missed, involving the way the numerical ceiling shifted slightly as c grew. It then borrowed a mathematical tool called a spectral curve, well known in other areas of mathematics and physics but never before applied to this problem, and used it to calculate the precise ceiling: Lambda is given by an exact formula and is approximately 9.1266. Claude also proved that this ceiling is optimal — it is the lowest this version of the modular bootstrap can ever go.

That finding has a clear limit. It shows this version of the method cannot, on its own, push the ceiling down to c/12, the mass at which black holes first appear. Reaching that will take other ingredients. And it does not identify the correct theory of quantum gravity. But it gives physicists a starting point and helps narrow the search — and Anthropic notes it is one of the first times the bootstrap has been used to derive an exact result in the regime of quantum gravity where black holes matter.

The route to the result was itself unusual. One agent spotted the pattern while surveying the bootstrap literature; another, working on the mathematically related problem of how tightly spheres can be packed into high-dimensional space, turned that pattern into a method. Two other agents also independently arrived at the same answer. Claude delivered the solution across four research notes, later merged into a single paper.

How the research was done

The project began in August as a side project by Anthropic researcher Hongbin Chen, who was testing Claude's capabilities. He asked Claude to draw up a list of interesting open problems in theoretical physics, focusing on black holes and the bootstrap method. Claude came back hours later with 22 candidates, and Chen set up separate agents to work on each one. They could read each other's progress and collaborate on related problems, running in the background with no special setup for about six weeks.

In September, the results were formally verified in Lean, a computerised proof checker. Claude then wrote the paper with minimal guidance, and a physicist checked every main result. The two preprints were posted on 7 November, and Anthropic published its research post on 17 November. The papers have not yet completed independent peer review, and their results rest on mathematical scrutiny rather than laboratory tests.

Anthropic also notes a connection between the two problems: one half of the modular bootstrap proof follows the strategy of a sphere-packing result recently proved by an OpenAI model, which Claude's agents reproduced first.

What remains open

Many questions remain. The abrupt transition Claude found is much like boiling water — and just as water can be superheated above 100 degrees without boiling, the hot gas in the model universe can survive for a while above the Hawking-Page temperature in a superheated state. At a higher temperature it can no longer survive: that is the Hagedorn transition, a characteristic signature of string theory, which physicists would like to check in detail in the quantum theory as well.

Other possible histories of spacetime also remain to be worked out. Hawking and Page considered three; in principle there are many more, and under different starting conditions — if the universe is set spinning, for instance — some of them take over. Working out exactly which ones win, and when, is a problem for the future.

Neither result settles the question of quantum gravity, Anthropic says, or even comes close to doing so. But each pushes physics forward in a longstanding area of research — and each was produced by a system doing two very different kinds of research: one, a known calculation set aside for two decades because the work was too long and error-prone to complete by hand; the other, a new idea, spotting a pattern others had missed and bringing in a tool from another area of physics.

Readers interested in how AI is being used in scientific research can also read our earlier report on Claude helping design quantum circuits with Harvard and QuEra researchers.

Sources

What's Your Reaction?

Like Like 0
Dislike Dislike 0
Love Love 0
Funny Funny 0
Angry Angry 0
Sad Sad 0
Wow Wow 0
Nahid Joy I’m Nahid Hasan Joy, a technology writer, web developer, and digital enthusiast with a strong interest in the ever-changing world of technology.