Anthropic says Alibaba must be punished for largest Claude cloning attack

The scale here matters more than the act itself: 25,000 coordinated accounts across 29 million interactions suggests this was an organized, resource-intensive operation rather than opportunistic scraping, which means Anthropic's legal theory will likely rest on proving intent and coordination rather than simple ToS violation.

Modelwire has no prior coverage directly connected to this case. It belongs to a broader pattern of model IP disputes that has been building across the industry as frontier labs monetize API access and well-resourced competitors seek shortcuts around the training compute required to match frontier performance. The legal infrastructure for adjudicating these disputes is still largely untested, and this case may become a reference point for how courts treat systematic output harvesting as a form of misappropriation. What Anthropic is really stress-testing is whether contractual remedies, rather than technical countermeasures, can serve as a credible deterrent against extraction at this scale.

Watch whether Anthropic wins any injunctive relief before the case reaches discovery. A preliminary injunction would signal courts are willing to treat coordinated extraction as immediately actionable harm, which would materially change how every other frontier lab structures its API enforcement posture.