Anthropic: You can still use your Claude accounts to run OpenClaw, NanoClaw and Co.

Anthropic clarifies its stance on personal AI agents

Anthropic says a recent update to its terms of service documentation was a mistake and that users can still run popular personal AI agents like OpenClaw and NanoClaw with their Claude subscriptions. The company sparked confusion last week when it updated its Claude Code documentation to state that using OAuth tokens from consumer accounts in other products was against its terms.

The new language specifically warned developers building products that interact with Claude to use official API keys instead. “Anthropic does not permit third-party developers to offer Claude.ai login or to route requests through Free, Pro, or Max plan credentials on behalf of their users,” the updated document read.

Community backlash and immediate clarification

This caused immediate concern on developer forums like Reddit and X. Many developers use the flat-rate $20/month Claude Pro plan to power coding agents and experimental tools via the company's own Agent SDK, avoiding more expensive pay-as-you-go API costs.

Anthropic staff moved quickly to clarify the situation. Thariq Shihipar, who works on Claude Code, posted on X that the change was a “docs clean up” that caused confusion. “Nothing is changing about how you can use the Agent SDK and MAX subscriptions!” he wrote.

He later added that Anthropic wants to encourage experimentation, but “if you’re building a business on top of the Agent SDK, you should use an API key instead.” The company confirmed to The Verge that it will not be canceling user accounts over this issue.

A history of mixed signals

Despite the reassurance, the incident has heightened developer anxiety. This is partly due to other recent actions from Anthropic that have been perceived as restrictive. The company has previously:

• Asked the founder of “Claudebot” to change the project’s name, leading to its rebrand as OpenClaw.
• Banned developer tools like OpenCode from using its OAuth system for authentication.

These moves, combined with the ambiguous documentation update, have led many in the community to believe the current era of affordable, flat-rate AI access is ending. There's a widespread feeling that consumer subscriptions are subsidized by higher-priced enterprise API usage.

The future of AI agent development

The core tension lies in distinguishing between personal experimentation and commercial product development. Anthropic’s official statement attempts to draw this line, affirming support for the former while steering the latter toward its formal API.

“The update was a clarification of existing language in our docs to make it consistent across pages,” a company spokesperson said. However, the episode underscores the precarious position of developers building on top of major AI platforms, where policy changes can happen with little warning.

For now, projects like OpenClaw and NanoClaw can continue operating as before. But the confusion has made it clear that the rules of engagement for the AI agent ecosystem are still being written, and developers are watching every move.