Recently, the renowned AI company Anthropic has revised its terms of service, expressly forbidding users from linking their Claude subscription accounts to third-party automated AI tools, such as OpenClaw, through OAuth authentication. In the past, certain tech enthusiasts leveraged monthly subscription accounts and OAuth authorization to allow OpenClaw to carry out tasks. This led to significant consumption of computing resources and exerted cost pressures on Anthropic. In reaction, Anthropic has implemented measures to suspend accounts that misuse the interface and stressed that only the official Claude Code and web-based interface are permitted to legally utilize OAuth tokens. Should users persist in using OpenClaw, they will be required to transition to the official API key model, which operates on a pay-as-you-go pricing structure. This alteration in policy underscores the clash over computing costs between large language model service providers and open-source agent tools. As a result, the cost of compliant usage for developers is set to increase, and the pursuit of localized large model solutions may emerge as a new trend.