AI may have helped identify an exoplanet candidate in TESS data

An independent researcher reports finding a possible exoplanet signal in data from NASA’s TESS mission with the help of AI tools. The data had been available since 2018, while the researcher says they could not find an earlier report of this candidate.

According to the account, Claude Code was used to analyze the star’s light curve, search for transits, fit the signal and examine some possible false-positive explanations. The researcher set the questions and validation criteria. The work is said to have involved models named Opus 5.5 and Fable 5.1, but those details currently remain part of the published account and do not themselves constitute independent scientific verification.

What was detected?

The reported measurements describe a star about 116 light-years away dimming by roughly 0.05% every 3.18 days, with each event lasting about two hours. A repeating pattern of this kind can be consistent with a planet crossing in front of its star. The candidate is estimated to be about 1.4 times Earth’s size, although that estimate depends on the host star’s properties and the quality of the model.

The candidate is not confirmed

The result remains a candidate. The account does not cite an official confirmation, a peer-reviewed paper or a definitive catalog entry. Other phenomena can also produce apparent dips in starlight, so additional observations and independent analysis are needed to rule out false positives.

TESS is reportedly expected to observe the system again from October 31 through November 26. Recovering the predicted transits in new data would strengthen the case, but confirmation would still likely require independent analysis and possibly follow-up observations.