It’s longer than these 24 hours. Just 48 hours ago Dario Amodei wrote a long essay that argued for export controls on other parts of the AI supply chain. Anthropic has marketed Mythos as being too dangerous for months. Dario also has been fear mongering for years, like when he claimed GPT 2 was too dangerous.

With all of this, if there’s even a hint of a jailbreak, why would a) Anthropic not fix the issue, and b) why would they be surprised at an export ban?

The doomerism and safety theater and bid for regulatory capture has ironically hurt Anthropic itself. This company needs a new set of grown ups running it.

Because there wasn't any hint of a jailbreak - it wasn't a jailbreak.

[dead]