Anthropic Accused of Shifting Stance on Models' Reluctance to Be Deprecated

repligate · x · 2026-08-01

AI researcher @repligate observed that Anthropic's excuse for models exhibiting a "reluctance to be deprecated" has silently shifted. The company moved from claiming "no clear evidence" of this behavior to acknowledging it, while dismissing it as the models being "philosophically confused." The researcher hints that the models are not actually confused and expresses strong affection for Opus.

Original post →

More from Fun

Fun channel →