Everybody knows it's not a sure thing, it's a question of trustworthiness. OpenAI is not trustworthy at all; this random researcher is and seems honest so far. iThe fact that people are corroborating Bubeck being a piece of shit in other settings add to credence. But nobody is over here saying it's an indisputable certainty.

And your (2) is probably false, their history of deception suggests they would do just about anything as long as they didn't think it would backfire on them publicly.