This is a special post for quick takes by Tymoteusz. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
Mythos has been released on the day of it's launch, and I haven't seen any discussion about it--- It really looks like Anthropic is dropping the ball on it's Infosec.
Bloomberg broke the story here
- A contractor got access to the model on the day of release to project glasswing partners
"To access Mythos, the group of users made an educated guess about the model’s online location. They based this on knowledge about the format Anthropic has used for other models, the person said, adding that such formatting details were revealed in a recent data breach from Mercor, an AI training startup that works with a number of top developers."
The contractor was originally doing evals for Anthropic
"Crucially, the person also has permission to access Anthropic models and software related to evaluating the technology for the startup. They gained this access from a company for which they have performed contract work evaluating Anthropic’s AI models. Bloomberg is not naming the company for security reasons."
This reads to be a similar mistake that was done about the URL of the unreleased blogposts that mentioned Mythos.