Anthropic details distillation campaigns from Alibaba, Moonshot AI, and DeepSeek
A new report released Thursday by Anthropic alleges persistent distillation attacks by China-based AI companies, which have escalated in recent months as competition in the space has intensified.
TL;DR
- Anthropic's new report alleges escalating distillation attacks from China-based AI companies.
- These attacks aim to harvest capabilities and internal 'chain of thought' from US frontier AI models.
- Anthropic observed nearly 200 million exchanges linked to five separate distillation attack campaigns.
- A campaign attributed to Alibaba involved 151 million exchanges used to train its Qwen models.
- Another campaign from Moonshot AI allegedly routed requests from the Chinese military, targeting Claude's Opus model.