"Released" is a fuzzy term. If you mean published, well, yes, they should not be published to a peer reviewed publication, but most peer review publications wouldn't bother publishing them in the first case. Preliminary results/progress reports can and probably should be released in many cases, if only to let people know what's going on. That happens a lot, you get "Oh we noticed something with three sigma significance over here, but we don't have enough statistics" and then somebody else can examine their data and say "well I've got five sigma proof that's an artifact" and then more papers are written on methods, cuts, calibration, simulation, etc... while both experiments might still be running.
On one hand, sometimes this act itself can bias people into making bad cuts and stuff. On the other hand, when this process doesn't happen, you also get people working in a box for two years with some systemic error they didn't account for and then claiming a 6 sigma feature in publication that's really an artifact of noisy ADC channel and some massaged cuts.