OpenAI Accuses Moonshot AI of Mass Data Extraction, 16,000 Daily Requests
SingTao · 3 SOURCESabout 2 hours ago2 MIN
Summary
OpenAI has accused Chinese AI company Moonshot AI of conducting a coordinated campaign to extract proprietary reasoning data from its GPT systems, with the operation peaking at 16,000 requests in a single day during late July . The attack, described as 'adversarial distillation,' sought to capture hidden inference processes that reveal how AI models solve problems . OpenAI detected the activity across more than 15,000 users and blocked it on July 28, warning that such extraction poses potential security and national security risks . This accusation follows similar claims by Anthropic last month and comes as Moonshot's Kimi K3 model has gained attention for rapidly closing the gap with American AI systems .
Key Points
- The coordinated attack began in early July, with requests surging to 16,000 on July 24–25, involving over 4,000 users
- OpenAI identified activity across more than 15,000 users in total and successfully blocked the operation on July 28
- Attackers manipulated user interactions to reproduce hidden reasoning processes rather than breaching encryption or databases
- Anthropic had previously accused Moonshot of redirecting thousands of user requests to its Claude model to train Kimi
- OpenAI's Caroline Zier emphasized the company targets service violations, not open-source models or legitimate distillation
Why It Matters
The case highlights escalating tensions in the global AI arms race, where companies seek to replicate advanced capabilities at minimal cost through data extraction. For Hong Kong, this dispute underscores the growing intersection of AI development with geopolitical competition, as both US and Chinese firms compete for technological supremacy that could reshape industries from finance to healthcare. The allegations also raise questions about the adequacy of current intellectual property protections in the rapidly evolving AI sector.
The case highlights escalating tensions in the global AI arms race, where companies seek to replicate advanced capabilities at minimal cost through data extraction. For Hong Kong, this dispute underscores the growing intersection of AI development with geopolitical competition, as both US and Chinese firms compete for technological supremacy that could reshape industries from finance to healthcare. The allegations also raise questions about the adequacy of current intellectual property protections in the rapidly evolving AI sector.