Anthropic reveals Claude Mythos preview breaks simplified AES encryption
Technology
Anthropic just revealed that its new AI, Claude Mythos Preview, found a clever way to break into a simplified version of the Advanced Encryption Standard, or AES, the technology that helps keep your online payments and messages safe.
The announcement dropped on July 29, 2026, and it's got people talking about how smart AI is getting.
Anthropic calls for global rules
Claude's discovery means it can crack this weaker version of AES between 200 and 1,000 times faster than humans could before.
While the AES we actually use isn't at risk yet, the news highlights how quickly AI is learning to spot, and possibly exploit, flaws in important systems.
Anthropic says this is a wake-up call for stronger rules and global teamwork to make sure AI stays safe as it evolves.