"Dario never had a moat": debate claims Claude's RL builds on OpenAI and DeepSeek work

AestherML · x · 2026-09-06

AestherML sparked debate arguing Anthropic never had a moat: OpenAI was always ahead and dishonest about models served to the public, and most of the RL research underlying Claude builds on OpenAI and DeepSeek work — leaving mechanistic interpretability as Anthropic's main distinct achievement.

Original post →

More from Companies & People

Companies & People channel →