Anthropic blocks Claude subscription use in OpenClaw

Anthropic is ending the "infinite loop" subscription hack to protect margins and push developers toward their first-party agent ecosystem. This change effectively turns a $20/month flat fee into hundreds or thousands of dollars for heavy users of autonomous agents. Anthropic claims subscription tiers were never intended for high-volume "looping" usage, which places disproportionate stress on compute resources. The timing is critical as OpenClaw recently reached 100k GitHub stars and a top Product Hunt ranking, threatening Anthropic's control over agentic workflows. Developers view this as a strategic moat-building move to favor Anthropic’s own "Claude Code" and "Dispatch" features over third-party alternatives. The "claw tax" highlights the growing tension between personal chat subscriptions and the resource-heavy reality of persistent AI agents.