The academic peer review papers say something a lot different than the catchy headlines. "The measured signatures of thrust are subtle experimental errors. Possible sources include thermal effects, problems with magnetic shielding, or even a non-uniform gravitational field."
Nothing new then, this was the conclusion several months ago; but it's good to see it accessibly written out on a popular news site.
It is almost certainly a methodological error in the experiment; but that fact that so many groups are experiencing the same error (Even NASA's JPL, of all groups) implies that there's some thing unknown to science causing the error, even if it's a failure in the testing apparatus. Some of the best minds in their fields can't yet work out what's causing the effect, and that means we may well learn something new – all be it something that agrees with our current understanding of momentum conservation and thermodynamics.