Anthropic has temporarily banned the creator of OpenCL from accessing the cloud
Early Friday morning, OpenCL creator Peter Steinberger announced at X that it will likely become more difficult to maintain compatibility between OpenCL and anthropic models in the future. His post included a screenshot of a notification from Anthropic stating that his account had been suspended due to ‘suspicious’ activity.”
“The suspension proved temporary; within hours of his post going viral, Steinberger announced that his account had been reinstated. The comment section was quickly flooded with conspiracy theories, likely fueled by the fact that Steinberger is currently employed by Anthropic’s rival, OpenAI. Amid the speculation, an Anthropic engineer intervened to clarify that the company has never banned users for using OpenClaw and offered to assist the developer.”
It is still uncertain whether that conversation was the deciding factor in the account’s recovery, and we have reached out to Anthropic for clarification. However, the entire conversation provided important insight into the situation.
“To provide some context: this suspension followed an announcement last week from Anthropic, stating that its Claude subscription plans would no longer support ‘third-party harnesses,’ specifically mentioning OpenClaw.”
“OpenClaw users are now required to pay for their usage on a consumption basis directly through Claude’s API. Because Anthropic competes with OpenClaw through its own agent, Cowork, this shift is effectively functioning as a ‘claw tax.’ Steinberger maintains that he was adhering to these new requirements and utilizing the API as instructed, yet he was suspended nonetheless.”
“Anthropic stated that it implemented the pricing shift because existing subscription plans were not designed to accommodate the unique ‘usage patterns’ associated with claws. Unlike standard prompts or basic scripts, claws are far more compute-intensive, often executing continuous reasoning loops, automating task retries, and integrating extensively with third-party tools.”
“Steinberger, however, remained skeptical of Anthropic’s explanation. Following the pricing adjustment, he posted, ‘Funny how timings match up, first they copy some popular features into their closed harness, then they lock out open source.’ While he did not explicitly name the features, his comment likely alluded to the recent update to Claude’s Cowork agent, ‘Claude Dispatch,’ which allows users to remotely control agents and assign tasks. Notably, Dispatch was released just two weeks prior to Anthropic’s shift in OpenClaw pricing.”
Steinberger’s ongoing tensions with Anthropic remained evident on Friday. When a social media user suggested that his current predicament was a result of his choice to join OpenAI over Anthropic, posting, “You had the choice, but you went to “Wrong people,” Steinberger countered: “One welcomed me, another sent legal threats.”
That is a stinging reply. When others questioned why he would use Claude at all given his role at a competitor, Steinberger clarified that his usage is strictly for testing, ensuring that updates to OpenClaw do not negatively impact those who use the tool with Claude.
“You have to separate the two issues,” he clarified. “One is my work at the OpenCL Foundation, where we want to make sure that OpenCL works great for any model provider; and the other is my job at OpenAI, where I help create future product strategy.”
Claude’s continued preference among OpenClaw users—over ChatGPT—was cited by many as the reason testing remains essential. Addressing Anthropic’s pricing shift, Steinberger replied, “Working on that,” which may signal the nature of his current responsibilities at OpenAI. He did not respond to requests for comment.
