Anthropic temporarily blocked Peter Steinberger, the developer behind OpenClaw, from accessing its Claude AI models. The suspension occurred early Friday morning, with Steinberger posting a screenshot on X showing a message from Anthropic citing “suspicious” activity as the reason for his account ban.
The restriction did not last long. Within hours, after Steinberger’s post gained widespread attention, his account was restored. An Anthropic engineer commented on the thread, stating that the company has never banned anyone for using OpenClaw and offering assistance. It remains unclear whether this intervention directly led to the reinstatement.
This incident followed an announcement from Anthropic the previous week. The company declared that subscriptions to Claude would no longer cover usage through “third-party harnesses including OpenClaw.” Instead, OpenClaw users must now pay separately via Claude’s API based on their consumption levels.
Anthropic explained the pricing adjustment by pointing to the unique demands of tools like OpenClaw. These applications often involve more compute-intensive operations, such as continuous reasoning loops, automated task retries, and integrations with external services. The company argued that its subscription plans were not designed to handle such usage patterns effectively.
Steinberger, however, rejected this justification. He claimed to be adhering to the new API billing rules when his account was suspended. In a previous post, he suggested a strategic motive behind the timing, noting, “Funny how timings match up, first they copy some popular features into their closed harness, then they lock out open source.”
He likely referred to features like Claude Dispatch, which allows remote control and task assignment for agents. This functionality launched a couple of weeks before Anthropic revised its OpenClaw pricing policy.
The ban reignited public criticism from Steinberger. When one commenter implied he faced repercussions for joining OpenAI instead of Anthropic, Steinberger retorted, “One welcomed me, one sent legal threats.”
Others questioned why he continued to use Claude given his employment at OpenAI. Steinberger clarified that his usage was strictly for testing purposes, to ensure OpenClaw updates remain compatible with Claude for its users. He emphasized a distinction between his role at the OpenClaw Foundation, focused on broad model provider support, and his position at OpenAI, where he contributes to product strategy.
Commenters noted that Claude remains a preferred choice for many OpenClaw users over alternatives like ChatGPT, necessitating ongoing compatibility testing. When asked about adapting to Anthropic’s new pricing, Steinberger responded, “Working on that,” hinting at potential developments in his work at OpenAI.
Steinberger did not provide additional comments when contacted for this article.
