AI Distillation Attacks on Claude Infrastructure (infosecurity-magazine.com)

• DeepSeek, Moonshot, and MiniMax identified as primary actors.
• Systematic use of model outputs to train rival domestic architectures.
• Highlights vulnerabilities in API-based AI delivery models.

"Anthropic has formally accused Chinese AI firms DeepSeek, Moonshot, and MiniMax of conducting 'distillation attacks' against its Claude model. These techniques involve querying the model at scale to extract its internal logic and decision-making patterns, effectively stealing intellectual property to enhance their own competitive AI systems."