[ta988]

You shouldn't "trust" papers, stay critical and verify. Wherever they come from. There is a lot of politics, grad students eager to graduate so they cut corners, cheating PIs, cheatings statisticians... (I've witnessed each of these during my career). What you should trust is when things get built upon other works (from other groups) or when it simply gets reproduced. This does not eliminate the risk of fraud or error but greatly reduce it. The same way do not trust claims from companies based on a single paper especially if the company is run by one of the authors. Again it is just my limited experience but most of the ones I have seen were just full of overblown claims and they just hoped they could jump the ship before it got discovered.