Anthropic temporarily suspended the account of Peter Steinberger, the creator of the OpenClaw project, citing suspicious activity. The incident began early Friday when Steinberger announced the ban on X, attaching a screenshot of the suspension notice. However, the restriction was short-lived, as his access was restored a few hours later following the viral nature of his post. The controversy stems from recent policy changes by Anthropic regarding its Claude models. Last week, the company announced that its standard subscriptions would no longer cover third-party automation tools known as "claws," including OpenClaw. Instead, users must now pay for these extensive usage patterns separately through the API based on consumption. Anthropic described this shift as a response to the unique demands of claws, which perform continuous reasoning loops and heavy compute tasks that exceed the scope of simple prompts. Effectively, this introduced a new charge for automation usage. Steinberger claimed he had already complied with the new pricing structure by using the API, yet he was still banned. He expressed skepticism about Anthropic's justification, noting the timing coincided with the launch of Claude's proprietary agent, Cowork, which includes features similar to those found in OpenClaw. He suggested a pattern of copying popular open-source features into closed offerings before restricting access to the original open implementations. The situation is complicated by Steinberger's employment status. He recently joined OpenAI, a direct competitor to Anthropic, which has fueled speculation and conspiracy theories among online observers regarding the ban. An Anthropic engineer publicly commented on Steinberger's post, stating that the company has never banned users for using OpenClaw and offered assistance. It remains unclear whether this intervention led to his reinstatement. Despite working for OpenAI, Steinberger explained his continued use of Claude models for testing purposes. He emphasized the distinction between his role at the OpenClaw Foundation, which aims to ensure compatibility across all model providers, and his job at OpenAI, where he helps shape future product strategy. He noted that testing against Claude is essential because it remains a popular choice for OpenClaw users. When asked about the pricing change, he indicated he is addressing the issue, suggesting that his work at OpenAI involves negotiating or resolving such compatibility and cost disputes. This brief suspension highlights the growing friction between AI developers building open automation tools and model providers adjusting their monetization strategies. It also underscores the complexities of the current AI landscape, where former employees and industry insiders navigate competing interests while maintaining their commitment to interoperability. Steinberger did not provide further comment on the matter.