Peter Steinberger, the creator of OpenClaw, publicly expressed concerns early Friday morning on X, stating, "it’s gonna be harder in the future to ensure OpenClaw still works with Anthropic models." His post included an image indicating his Anthropic account had been temporarily suspended due to "suspicious" activity.
The suspension was brief. Within hours, following the viral spread of his post, Steinberger confirmed his account had been reinstated. Amidst numerous comments, many speculating conspiratorially given Steinberger's employment with Anthropic's competitor, OpenAI, an Anthropic engineer weighed in. The engineer assured the renowned developer that Anthropic does not ban users for utilizing OpenClaw and extended an offer of assistance.
It remains unconfirmed whether this intervention directly led to the account's restoration. Nonetheless, the entire public discourse offered significant insights into the ongoing dynamics between AI developers and platform providers.
This incident occurred shortly after Anthropic announced a significant policy change the previous week: subscriptions to its Claude AI model would no longer cover "third-party harnesses including OpenClaw."
Consequently, OpenClaw users are now required to pay for their usage separately, based on consumption, directly through Claude’s API. This shift effectively introduces a "claw tax" by Anthropic, which also offers its proprietary agent, Cowork. Steinberger asserted that he had complied with this new payment structure by using his API, yet his account was still suspended.
Anthropic justified the pricing alteration by explaining that its subscription model was not designed to accommodate the "usage patterns" characteristic of "claws." These tools can be more compute-intensive than standard prompts or simple scripts, often involving continuous reasoning loops, automated task repetition or retries, and integration with various third-party applications.
Steinberger, however, remained unconvinced by Anthropic's explanation. Following the pricing change, he posted, "Funny how timings match up, first they copy some popular features into their closed harness, then they lock out open source." While not explicitly named, this comment was likely a reference to features integrated into Claude's agent Cowork, such as Claude Dispatch, which facilitates remote agent control and task assignment. Dispatch had been rolled out just weeks before Anthropic revised its OpenClaw pricing policy.
Steinberger's palpable frustration with Anthropic was evident once more on Friday.
In one exchange, a commenter suggested Steinberger shared responsibility for the situation, having chosen to work for OpenAI instead of Anthropic, stating, "You had the choice, but you went to the wrong one." Steinberger's sharp retort was: "One welcomed me, one sent legal threats."
When questioned by multiple individuals about his continued use of Claude models over those of his employer, he clarified that his usage is strictly for testing. This ensures that OpenClaw updates maintain compatibility and functionality for its Claude users.
He further elaborated on his distinct roles: "You need to separate two things. My work at the OpenClaw Foundation where we wanna make OpenClaw work great for *any* model provider, and my job at OpenAI to help them with future product strategy."
Several commenters also underscored that the necessity to test Claude stems from its enduring popularity among OpenClaw users, often preferred over ChatGPT. Responding to Anthropic's pricing changes and the continued user preference for Claude, Steinberger remarked, "Working on that," hinting at a potential project within his OpenAI responsibilities.
Steinberger did not provide further comment when contacted.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.