A perfectly rational response. Security-by-obscurity always works. It's one of the best #security methods out there. There's absolutely no way that an entity outside of the US's jurisdiction doesn't eventually come up with a Mythos-equivalent that isn't subject to US embargo.

Whatever your opinions of
#AI, this doesn't seem to be the smartest move. Feels like all it does is prevents projects from being more proactive about security. Definitely a Good Idea™.

#Mythos
If even someone as inexpert as me sees the flaws in the Federal government's approach to the "problem" that #Mythhos presents, it's a near inevitability that people with larger platforms and greater focus and expertise will point them out, as well.

Shit like this is why you don't gut
#CISA and your other guardians.

#AI
#Security
#AmateurHour
No failover for the Commerce Secretary

The US government switched off Anthropic's two best models for every customer on Friday evening. Your disaster-recovery plan has a runbook for an outage, yet nothing for a letter from Commerce.

Artificial Confidence