AI & Models
Anthropic reinstates OpenClaw creator after account ban
Anthropic temporarily suspended OpenClaw creator Peter Steinberger, highlighting growing friction between AI model providers and third-party developers over pricing and access to model harnesses.
On early Friday morning, OpenClaw creator Peter Steinberger posted on social media about his account suspension by Anthropic, which cited suspicious activity. Steinberger shared an image of the suspension notice and stated that it would become more difficult in the future to ensure OpenClaw remains compatible with Anthropic’s models. The post generated hundreds of comments online. Following the viral response, Anthropic reinstated Steinberger’s account. It is not clear if the restoration was influenced by a public comment from an Anthropic engineer, who stated that the company has never banned anyone for using OpenClaw and offered to assist.
The suspension follows news last week that subscriptions to Claude, Anthropic’s AI model, would no longer cover third-party harnesses—external tools or software that interface with AI models—including OpenClaw. Under the new policy, users must pay for this usage separately through Claude’s API, a pricing shift colloquially referred to as a claw tax. Steinberger, who was using his API under the new rules when he was suspended, suggested the pricing change may be linked to features added to Cowork, Anthropic’s agent product. Commenting on the timing, Steinberger suggested that the company copied popular features into its closed harness before locking out open-source alternatives.
The situation is further complicated by Steinberger’s employment at OpenAI, a rival to Anthropic. Addressing the relationship between his two roles, Steinberger noted that he has faced legal threats from Anthropic. Peter Steinberger, the creator of OpenClaw, contrasted his experiences with the two companies, stating, “One welcomed me, one sent legal threats.” When questioned about his continued use of Claude, Steinberger explained that he only uses the model for testing to ensure OpenClaw remains compatible for its users. He emphasized that his work at the OpenClaw Foundation, which aims to make the tool compatible with any model provider, must be separated from his role at OpenAI helping with future product strategy. When users noted that Claude remains a popular choice over ChatGPT for OpenClaw users, Steinberger replied that he was working on that.
Why it matters
The incident underscores the friction between AI labs and the developer ecosystem as companies like Anthropic move to monetize their agentic features and restrict third-party access to their models.