• Anthropic says it’s been targeted in massive distillation attacks Anthropic alleges Chinese AI companies DeepSeek, Moonshot and MiniMax made 24,000 accounts and 16 million Claude exchanges to scrape its AI bot for training. • Cointelegraph in your social feed Artificial intelligence firm Anthropic has accused three AI firms of illicitly using its large language model Claude to improve their own models in a technique known as a “distillation” attack. • In a blog post on Sunday, Anthropicsaidthat it had identified these “attacks” by DeepSeek, Moonshot, and MiniMax, which involve training a less capable model on the outputs of a stronger one. • Anthropic accused the trio of generating “over 16 million exchanges” combined with the firm’s Claude AI across “approximately 24,000 fraudulent accounts.” “Distillation is a widely used and legitimate training method. • For example, frontier AI labs routinely distill their own models to create smaller, cheaper versions for their customers,” Anthropic wrote, adding: Anthropic said that the attacks focused on scraping Claude for a wide range of purposes, includingagentic reasoning, coding and data analysis, rubric-based grading tasks, and computer vision. • “Each campaign targeted Claude’s most differentiated capabilities: agentic reasoning, tool use, and coding,” themulti-billion-dollar AIfirm said.

Article Summaries:

  • Anthropic has publicly accused three Chinese AI firms-DeepSeek, Moonshot, and MiniMax-of conducting large‑scale “distillation” attacks on its Claude language model. The company claims the attackers generated over 16 million exchanges with Claude through roughly 24,000 fraudulent accounts, targeting Claude’s advanced capabilities in agentic reasoning, coding, tool use, and data analysis. Anthropic identified the campaigns via IP correlation, request metadata, and industry partner observations. It warns that such illicit distillation could transfer powerful U.S. AI technology to foreign governments for military or surveillance use. The firm plans to strengthen detection, share threat intelligence, and urges industry‑policy collaboration to curb similar attacks.

Sources: