Anthropic restricts Claude subscribers from using OpenClaw, starts refunding users

Users can still access OpenClaw with Claude models, but only through additional paid usage bundles (now offered at a discount) or by using a full Claude API key, which follows pay-per-token pricing.

Anthropic is in the news once more, and this time, its decision has sparked significant backlash. The AI firm announced that Claude subscriptions will no longer cover usage on third-party tools like the popular open-source AI agent OpenClaw. From April 4, 2026, users must pay extra to access such tools, even if they already have a paid Claude Pro or Max subscription.

Anthropic executive Boris Cherny broke the news on X (formerly Twitter), explaining, "Starting tomorrow at 12 pm PT, Claude subscriptions will no longer cover usage on third-party tools like OpenClaw." He added that Anthropic's computing capacity was not optimised for the usage patterns of these tools, and subscriptions were not designed for them.

Users can still access OpenClaw with Claude models, but only through additional paid usage bundles (now offered at a discount) or by using a full Claude API key, which follows pay-per-token pricing.

Claude users hurt by Anthropic's decision

Anthropic's decision has triggered widespread frustration on social media. Many subscribers have expressed disappointment, with one user writing, "I love you guys, and Claude is amazing, but this low-key sucks. This decision is actually going to actively hurt a lot of people."

Another user questioned the rationale, "The increasing demand is nothing to do with third-party use. The amount of tokens is the same. Why do you treat us as stupid?"

OpenClaw's creator, Peter Steinberger, who recently joined OpenAI, revealed that he and others had tried to negotiate with Anthropic and were able to manage only a one-week delay. He hinted at timing concerns too, noting that Anthropic had recently added similar agentic features to Claude, such as computer control and task automation.

OpenClaw is an open-source agentic AI platform that allows users to run autonomous AI agents locally or on their devices. It gained massive popularity for enabling Claude models to perform complex, independent tasks such as automation, app building, and persistent assistance across messaging platforms.

Anthropic's response to the fiasco

In response to the backlash, Anthropic stated it is issuing full refunds and discounts for any overages caused by the change. Affected users are expected to receive an email with a link to claim refunds starting tomorrow.

The company has also been rolling out its own native agentic capabilities in Claude, including features like Dispatch for computer control and remote phone access. Some users see this as a direct competition to tools like OpenClaw.

This move comes amid surging demand for Claude, with several users recently reporting faster exhaustion of their usage limits. Anthropic maintains that no one was charged unfairly and has suggested ways for users to optimise token consumption.