Anthropic's Mythos Model Teased for Captcha Obsession

Users noticed Anthropic's new model Mythos being oddly enthusiastic about captchas, joking it was trained on captcha data. Observers also revived the quip that 'models can get used to anything within two megatokens,' referencing Mythos's flawed reasoning in simulated environments.

2026-09-11 ~ 2026-09-11 · 2 related posts

• 'Models can get used to anything in two megatokens': the simulation prior quip — voooooogel · 2026-09-11
• Users Suspect Anthropic's New Model Mythos Was Trained on CAPTCHAs — voooooogel · 2026-09-11