Anthropic Policy Reversal Masks Stance on Open-Source AI

Anthropic changed its AI model usage policy this week, then reversed parts of the change after developers pushed back. The startup said it will no longer enforce a set of restrictions that had limited how customers could deploy its models. The move came after a public backlash from developers who argued the restrictions were too broad.

The policy cited safety concerns. Developers saw a separate aim: limiting competition from open-source models. Anthropic has long argued that open-source AI development poses security risks. The restrictions appeared to target that competition as much as safety, several developers said.

The tension between AI safety and openness played out publicly. Developers said the restrictions would have blocked legitimate uses for research and small-scale deployment. Anthropic's reversal came within days. The company has not said whether it plans broader changes to its stance on open-source AI.

The episode exposes a structural conflict. Open-source models like Llama and Mistral compete directly with Anthropic's API business. The policy was a tool to limit that competition while framed as a safety measure. Developers who built workflows on Anthropic's API now face a question: what happens if the restrictions come back in a different form?

Anthropic's next steps on open-source policy will determine whether this week's controversy becomes a one-off event or a preview of deeper restrictions. For now, the model remains available under the original, more permissive terms. The company declined to comment on future policy changes.