Rumor: Anthropic's model may have cracked popular encryption algorithms, disclosed only to authorities

basedjensen · x · 2026-11-21

User @zephyrz9 claims to be "pretty sure Anthropic has cracked a few popular encryption algorithms" but would never release the results due to the chaos it would cause, worrying that open models could do the same within a year. Reposter @teortaxesTex responds that the capability has likely already been "released to the relevant authorities." Unverified speculation, but a notable discussion about frontier-model security capabilities and disclosure.

Related event: Anonymous Claim: Anthropic Has Broken Major Encryption Algorithms(3 posts)→

Original post →

More from Models

Models channel →