Burkov mocks Amodei's inconsistency: feared GPT-2 in 2019, ships frontier coding models now

burkov · x · 2026-09-16

ML author Andriy Burkov mocked Anthropic CEO Dario Amodei's shifting stance: in 2019 he claimed fear of releasing GPT-2; since then Anthropic shipped frontier coding models while suggesting half of human coders could be fired; yet in 2026 he again cites safety fears to withhold a new model and urges rivals with similar models not to release. Burkov notes the quotes are from memory.

Original post →

More from Fun

Fun channel →