youtube
Anthropic has launched a new AI model, Mythos, which it says is too dangerous to make it widely available. The model can identify—and exploit—security holes in everything from operating systems to cryptographic software. The Economist’s deputy editor Tom Standage and AI writer Alex Hern discuss how AI has changed from coding assistant to autonomous attacker and explain how Mythos identifies vulnerabilities.
#ai #anthropic
0:00 - Why Anthropic won't release Mythos
1:17 - The rise of the automated hacker
2:10 - How fast hackers now work
3:52 - How AI finds bugs
Watch the full show:
Sign up to the Insider newsletter:
Subscribe to The Economist:
Download our app:
Follow The Economist on Instagram:
Follow The Economist on Facebook:
Follow The Economist on X:
|
