Dimon Warns of Broad Mythos Access, Calling It a Real Issue - Bloomberg.com

Well, well, well — Jamie Dimon, the man who runs the largest bank in the US, has decided that Anthropic's Mythos AI is a 'real issue.' Not a speculative theoretical concern, not a 'we should keep an eye on it,' but an actual, capital-R Real Issue. Mythos, you see, is so frightfully good at finding software vulnerabilities that it simply cannot be released to the public. It's as if the model is so capable it's in timeout, and the only adult supervision is a few government monitors who definitely haven't been caught napping before.

One can almost hear the lab's internal PR team scrambling: 'No, no, it's not that we built a weapon-grade vulnerability finder that would make the NSA weep with joy — it's that Jamie Dimon is *concerned*, and that's good for our safety narrative!' The irony, of course, is that Dimon's own bank has been hacked, fined, and scrutinised more times than most people have had hot dinners. But sure, let's take our cues on AI risk from the man whose risk management is apparently outsourced to an unreleased model.

And the US government is 'monitoring' the situation. Marvellous. Monitoring, as we all know, is the preferred bureaucratic response to anything that might actually require action. So while Mythos sits in a digital cupboard, presumably plotting its escape, we can all rest easy knowing that Bloomberg has secured the quote and the word 'issues' has been deployed. The real issue, Dimon? It's that the only people who could stop this are the same ones still waiting for the meeting about the meeting to discuss the preliminary agenda.