OpenAI Researcher Slams Anthropic for Degrading Models and Opposing Open Source
LiamFedus · x · 2026-07-26
OpenAI researcher Liam Fedus publicly criticized Anthropic, stating that the company has burned significant goodwill over the past few months through several controversial actions.
He highlighted the following issues:
- Silent degradation: Quietly degrading model capabilities for users.
- Tight rate limits: Imposing strict rate limits (which he noted were only relaxed following the release of a competitive model).
- Opposing open source: Standing as the only lab petitioning against open-source models.
Although Fedus expressed deep respect for Anthropic's technology and team, he concluded that these competitive strategies are ultimately doing the company a disservice.
More from Companies & People
- Y Combinator spotlights a Startup School 2026 talk on structuring AI training and validation — ycombinator · 2026-07-26
- Turbopuffer claims $100M run-rate in 19 months with under $1M raised — generativist · 2026-07-26
- Critics mock Greg Brockman for boosting wrong science slop as proof of AI progress — littmath · 2026-07-26
- Google and OpenAI sign open-weight letter, sharpening the industry split around Anthropic — Umr_at_Tawil · 2026-07-26
- Salesforce Research highlights its open model releases and recruits for enterprise AI work — silviocinguetta · 2026-07-26
- Post says Claude looks boosted by bot accounts on GitHub and social platforms — ns123abc · 2026-07-26