For this to be a lie, it wouldn't just be Sam Altman lying, it requires the entirely of OpenAI and METR to also be in on it. Do you think they would all fabricate an incident, staking a huge portion of their reputation on it, so they could spook people? What is the incentive, there are way more straightforward ways to lie about AI capabilities than a complex loss of control incident which now makes them liable to damages against huggingface.