OpenAI and Anthropic reportedly sitting on many unpublished internal model results

MoonL88537 · x · 2026-09-06

Thomas Bloom notes it is now widely known that both OpenAI and Anthropic hold many results proved by their internal models that they have never released, alongside their big public announcements. He opens a discussion on whether this selective disclosure of frontier lab research is good or bad for the field.

Original post →

More from AGI Musings

AGI Musings channel →