The Nature black hole star paper was posted as a preprint 17 months earlier, and rivals dispute it

A paper published in Nature in August 2026 is being reported as the discovery of a black hole star, a new type of astrophysical object. The paper is more careful than that, it was publicly available for seventeen months before it appeared, and at least three rival explanations for the same observations are in the literature.

The paper is "A gas-enshrouded and gas-reddened black hole at cosmic dawn", by Rohan P. Naidu of the MIT Kavli Institute for Astrophysics and Space Research and 53 co-authors, in Nature volume 656, issue 8127, pages 329 to 333, with the digital object identifier 10.1038/s41586-026-10846-4. Popular Science reported it on 12 August 2026. The object is MoM-BH*-1, drawn from the James Webb Space Telescope programme GO-5224, a survey the team calls Mirage or Miracle.

The claim in the abstract is narrower than the headlines. The authors present a source seen 660 million years after the Big Bang that exhibits three unusual properties: among the largest hydrogen Balmer breaks reported at any redshift, broad multi peaked H beta emission, and Balmer line absorption in several transitions. They then write that they "model this source as an enshrouded black hole", that it "may provide evidence" of an early black hole embedded in dense gas, and that radiation from the black hole "seems to dominate almost all observed light". The operative sentence is the last one: "The redness of the black hole is due to gas, not dust, and scattering, not kinematics, gives rise to the complex line shapes and luminosities, black hole masses of these sources may therefore be overestimated by orders of magnitude."

That is the finding worth reporting. If the red colour of the little red dots seen by JWST comes from light scattering through a dense hydrogen envelope rather than from dust, then the standard correction astronomers apply for dust is wrong, and the black hole masses derived from broad emission lines are wrong with it. The paper is an argument about a measurement error of orders of magnitude, not an announcement of a new kind of star.

The seventeen month gap is a matter of record. The same work was posted to arXiv as 2503.16596 on 20 March 2025 under the title "A 'Black Hole Star' Reveals the Remarkable Gas-Enshrouded Hearts of the Little Red Dots", with the JWST prism spectra deposited on Zenodo. The physics did not change in August 2026. Peer review concluded, and the title dropped the phrase "black hole star" in favour of "gas-enshrouded and gas-reddened".

The term itself is being stretched. The 2025 preprint used it for a configuration rather than a new class of star: extremely dense, turbulent gas forming a dust free atmosphere around a supermassive black hole. Naidu's group returned to the same physics in June 2026, describing a pseudo photosphere that forms within a dense wind and hides the central engine behind a blackbody like continuum. The Nature abstract uses neither the phrase nor its abbreviation, and speaks only of an enshrouded black hole and of the little red dots.

Independent work supports the gas interpretation without settling it. Bingjie Wang and colleagues, in a preprint posted on 5 February 2026, report an absorption feature at about 1.4 micrometres in two of four little red dots at redshift about 2, matching the shape and wavelength of the water band seen in cool stars. Atmosphere models require temperatures below about 3,000 kelvin to reproduce it, which the authors say confirms "unambiguously" a cool, dense gas component, contributing 20 to 30 per cent of the emergent continuum, and demonstrates that the red continua of some little red dots are "intrinsic rather than dust-reddened". They also report that a composite model suggests a range of about 2,000 to 4,000 kelvin rather than the single blackbody some gas envelope models predict. Wendy Q. Sun and colleagues, posting on 28 January 2026, used 98 sources with NIRSpec prism spectra and found the host subtracted median stack shows a Balmer break more than twice as strong as in massive quiescent galaxies. Andrea Weibel and colleagues, on 15 June 2026, compiled 241 candidates dominated by such a central engine across redshifts about 1.7 to 9.3.

The disagreement is real. Devesh Nandal and Abraham Loeb argue the spectra are produced by primordial supermassive stars rather than accreting black holes, and report matching the H beta width of MoM-BH*-1 to within 4 per cent. John Chisholm and colleagues, on 17 February 2026, propose the little red dots are globular clusters in formation, the rest frame optical light coming from a short lived supermassive star. Naidu's own group has since moved: in a paper posted on 29 June 2026 they use an escape velocity argument to constrain the engine to less than 100,000 solar masses for the typical little red dot, and favour intermediate mass objects of roughly 1,000 to a million solar masses radiating at more than five times the Eddington limit, harbouring a supermassive star or an intermediate mass black hole rather than a supermassive one.

What is not known is which model is right. No observation yet distinguishes a pseudo photosphere around a black hole from a supermassive star, the two candidates the same team has entertained within four months of each other. Naidu told Popular Science that "our picture of this object is evolving very rapidly". That is not a caveat added by a reporter. It is the state of the field.