A complete 100g combibar is normally intentionally overweight by about a gram to ensure that each of the 100 individual pieces weighs at least a gram when it's broken up.
Even so, it doesn't seem impossible for a single piece to fall slightly under weight if all of its neighbours won more than their fair share of the bendy boundary during the separation process.
On the other hand, how sure are you about the weight? It's very tempting to believe a measured weight is true and correct just because it appears on a digital display of something you own, but unless you have calibrated them I wouldn't trust them to be accurate, and I can understand why a seller might not want to trust them either.
I have a super-cheap set of digital scales that reads down to 0.001g (up to a max of 50g) and it comes with a 50g calibration weight. Of course, I have to trust the calibration weight actually weighs 50g for this to be of any use, but it does weigh exactly the same as another 50g calibration weight I have, so I have a reasonable amount of faith in it
Is it bullion by post? I'm sure they'll (or whoever else it might be will) respond to the video appropriately anyway. If not, you could demonstrate weighing something of a known weight, or simply the other two pieces as a comparison. Even a scale that doesn't read accurately is generally still useful for showing something is relatively heavier than something else.