AI Models Can Crack Everything But CAPTCHAs

[Smiggles]

AI Models Can Crack Everything But CAPTCHAs

The only thing between us and total destruction is identifying which images have buses in them.

gizmodo.com

Frontier AI models threaten to upend cybersecurity as we know it, and there have been multiple incidents of models breaking containment during training and hacking systems without guidance or permission while avoiding human detection. Yet it seems there is still one thing that these extremely powerful pieces of technology simply cannot do: solve a CAPTCHA efficiently.

 

[Smugs]

Yet...

 

[FiestaST]

 

[EmmanuelGoldstein]

They cannot crack embedded code I wrote either... Gemini and Claude fail spectacularly at it.

 

[Benedict A55h0le not]

They cracked whale language with AI