This is a tricky science result to explain and, as often turns out to be the case, not quite as secure or solid as the press release recyclers report.
The original paper is by Lauer et al. (2021) and measures the cosmic optical background (COB) using images taken with New Horizons, which is way beyond the orbit of Pluto. The advantage over things like HST is that the images are not contaminated by light scattering from zodiacal dust. Most of the paper deals with how they remove various instrumental signatures, get a figure for the dark sky brightness (about 27 mag per square arcsec in the V-band) and then attempt to explain this by examining the "light budget" from various known sources.
There are 4 main known sources. Faint stars below the detection limit of New Horizons (which is only about $V\sim 19$); ditto for galaxies; there is scattered light from bright stars out side the field of view; and there is a diffuse scattered light from dust in the Galaxy. The genuine COB is their measured sky brightness minus the contribution from stars and scattered light from our Galaxy.
They make an estimate of these 4 components and the sum comes to less than their dark sky brightness, but only by around 2 error bars; so a weakly significant result.
Now it gets confusing. In the text they say that the integrated light from faint galaxies (IGL) accounts for 50% of the COB, and may be consistent with it at the 2 error bar level. See Fig 18 in the paper.
Previous studies, based on HST data, suggested that if you extrapolated galaxy counts down to galaxies of about $10^6 M_\odot$, then there would be ten times as many unseen galaxies as you could count in the Hubble deep fields, mainly at $V>30$. However, if you work out how much IGL these would contribute it turns out to be larger than the COB that Lauer et al. have measured. Hence the headline that there may be fewer galaxies in the universe than previously claimed. I say may, because this claim is not one of the main conclusions of the paper, and in the text they say there may not be a significant discrepancy once the uncertainties are considered. This aspect of the work has been highly exaggerated in the press releases.
Lauer et al.'s estimate of the IGL is based on adding up the galaxies that have been observed down to $V\sim 30$. So you do need some extra faint galaxies to boost the IGL enough to fully explain the COB, but just hiding these at magnitudes $V>30$ may not be sufficient. The suggestion is that galaxies with $V<30$ are being undercounted. This is the origin of the second part of your report. It is that there may be more galaxies at brightnesses that ought to have been seen in the deep HST images. The implications of that are not explored.
Personally, I would like to take a look at how they have treated faint Galactic stars. Extrapolating from local luminosity functions may not be appropriate for estimating the contribution of stars down to $V=30$ in the halo of our Galaxy. If there are more lower mass objects in the halo than locally then this might have a bearing on the (2 sigma) discrepancy between the (apparent) COB and the IGL.