I’ve heard a lot about Mythos over the past few days. This article raises some questions about Anthropic’s claims.

These claims could be legitimate but it wouldn’t be the first time an AI company made bold statements a model’s performance that were found to be flawed after a period of time. The AI industry is moving so fast it is difficult to keep up at times but stories like this show that taking a step back is valuable.

Mythos sounds impressive and may very well be a major step forward. I don’t know that for sure. All I have to go on are scare stories coming from Anthropic and Project Glasswing. I won’t be able to tell for sure until it has been released and I can test it out.