Anthropic Bans OpenClaw Use via Subscription, Adds Fees

Anthropic, the AI safety company behind Claude, is implementing a significant policy change that restricts how developers can use its API. Effective April 4th, the company will no longer allow users to apply their standard Claude subscription credits or limits to third-party "harness" platforms like OpenClaw. Instead, access to Claude via these external agent-building tools will require separate, additional payments directly to Anthropic. This move effectively bans the use of OpenClaw under existing subscription plans and centralizes control over the burgeoning ecosystem of AI agents built on Claude. Developers who relied on these harnesses to create specialized workflows, automations, or complex applications will now face higher costs and must navigate a new payment structure. The policy shift highlights the growing pains in the AI platform economy. As companies like Anthropic invest heavily in developing powerful models, they are seeking to capture more value from the ecosystems that grow around them and ensure a consistent user experience. However, it also raises concerns about platform lock-in and could stifle innovation by making it more expensive for independent developers to experiment. For the AI community, this serves as a reminder that access to foundational models remains at the discretion of their creators, and the rules of engagement can change with a single policy update.