The new snippet says Anthropic’s prompt screening could create enterprise inefficiencies, raise costs and increase model lock-in risk through tighter control over outputs.
Chamath criticized Anthropic over allegations that it quietly downgraded models used by frontier research users. The updated snippet adds that Anthropic’s prompt screening could create enterprise inefficiencies, increase costs and deepen model lock-in risk by giving the company greater control over outputs. The material continues to frame the issue as a trust, innovation and compliance concern for enterprise AI users, but provides no timing, supporting evidence or response from Anthropic.