OpenAI has published a profile of astrophysicist Chi-kwan Chan, describing how he uses Codex, the company's AI coding tool, to build simulations of black holes. According to the profile, the simulations are intended to help scientists study extreme physics and test predictions of Einstein's theory of general relativity.
The piece is a single-source account from OpenAI itself, so it presents only Chan's experience and the company's framing of Codex's usefulness. There is no independent reporting or contrasting perspective in the source. The main claim is that Codex helps streamline the coding work involved in producing these simulations, but the article does not provide technical detail on the simulations or benchmarks.
Because there is only one source, the article should be read as a product story rather than a scientific finding. It does not claim that Codex made a new discovery, only that it assisted in the development of simulation code for studying black holes.