Peter Steinberger, creator of OpenClaw, announced that using OpenClaw with Anthropic models might become more difficult. He shared that his account had been suspended due to “suspicious” activity but was quickly reinstated after his post went viral. The brief suspension prompted comments, including one from an Anthropic engineer who offered assistance and clarified that using OpenClaw was not a bannable offense.
This incident follows the announcement that Anthropic’s Claude subscriptions would stop covering third-party harnesses, including OpenClaw, requiring users to pay based on usage through Claude’s API. Anthropic imposed this pricing change due to claws’ more intensive compute requirements compared to prompts or scripts. Steinberger complied with the new rule but was still banned.
Expressing skepticism, Steinberger suggested the timing of Anthropic’s policy changes correlated with introducing closed harness features similar to those in OpenClaw. His frustration with Anthropic was evident, responding to a comment about his job choice with OpenAI over Anthropic by stating that Anthropic had issued him legal threats.
Despite suggestions to use models from his employer, he clarified that he uses Claude for testing OpenClaw’s compatibility. Steinberger emphasized the separation between his role at the OpenClaw Foundation and his work at OpenAI, where he aids in future product strategy. Feedback about Anthropic’s pricing change reinforced Claude’s popularity among OpenClaw users, to which he indicated plans to address it. Steinberger declined to comment further.