OpenClaw (widely referred to as "Xiaolongxia" or "Crayfish"), an AI tool that generated significant buzz this year, has been prohibited by Anthropic. Previously, OpenClaw swiftly rose to prominence due to its remarkable efficiency in task handling. It enabled users to make high-frequency model calls at minimal costs via Claude subscriptions. Nevertheless, this model caused a cost reversal for Anthropic. Some heavy users were consuming far more computing power in a single day than what their subscription fees could offset.

On April 4, Anthropic declared a prohibition on third-party tools like OpenClaw from accessing models through Claude's subscription interface. This move compelled users to transition to more expensive API channels or buy extra data packages. The ban originated from long-standing disputes between the two entities. Initially named Clawdbot, OpenClaw was compelled to change its name due to its resemblance to Claude. After its founder, Peter Steinberger, joined OpenAI, Anthropic hastened the rollout of competing features (such as Claude Code/Cowork) and tightened its regulations.

Despite Anthropic providing one-time subsidies and discounted data packages, developers are still grappling with surging costs. For ordinary users, the monthly usage expenses have skyrocketed from $20 to thousands of dollars. This change signifies the AI industry's shift from "low-cost subscriptions" to "pay-as-you-go" models. Computing costs are now firmly on the rise, and ecological boundaries are reinforcing lock-in effects, which is a common business strategy to keep users within a specific ecosystem by making it difficult or costly to switch to competitors.