Anthropic has officially announced that, starting this week, Claude subscribers will no longer be able to access the AI service through third-party proxy tools such as OpenClaw. The move is intended to alleviate operational pressure on the platform during high-traffic periods and ensure the stability of core services.
Boris Cherny, head of Claude Code, stated clearly on social media: "As of 12:00 PM PT on Saturday, Claude subscription plans will no longer cover the use of third-party tools like OpenClaw."
Cherny explained that this restriction is driven by the need for engineering-level optimizations. He noted that Anthropic’s systems are currently highly tuned for specific workloads, and to ensure that as many users as possible can access their most intelligent models, the team must take steps to manage traffic. Despite this, he emphasized that Anthropic remains supportive of the open-source community, revealing that he recently contributed code to improve prompt caching efficiency for OpenClaw.
Addressing Frequent Service Disruptions
This adjustment comes at a time when Claude is facing significant stability challenges. On the Monday following the announcement, Claude.ai experienced a partial service outage. According to the company’s status page, users encountered numerous errors while logging in and using Claude Code and voice mode. Although the issue was resolved within an hour and a half that afternoon, the platform's uptime over the past 90 days has dropped to 98.82%.
Industry analysts believe this move by Anthropic is closely tied to cost control and adjustments to its business model. Currently, Anthropic offers services through both subscriptions and an API. Subscription users pay a flat fee for a set usage limit, while API users are billed per token. For high-frequency developers, the subscription model is often significantly cheaper than pay-as-you-go pricing.
Data suggests that some power users have been extracting token value far exceeding their $20 monthly subscription fee. As Anthropic prepares for a potential IPO, the company is looking to optimize its customer acquisition strategy to curb excessive resource consumption caused by low-cost subscriptions.
Anthropic is not the first AI company to take such action. In February, Google enforced stricter terms of service for Gemini CLI and Gemini Code Assist. At the time, Google pointed out that using third-party software to bypass OAuth authentication to scrape backend services was a direct violation of its usage policies.
For now, Anthropic has not announced whether it will offer specific API access solutions for third-party tools in the future. However, as the company moves toward its IPO goals, balancing user experience with infrastructure costs will remain a central operational challenge.
