Key Points
- Six Chinese AI developers, including Moonshot, DeepSeek and Alibaba, stand accused by U.S. intelligence of systematic model distillation.
- Federal agencies claim the companies harvested billions of tokens by sending millions of queries to GPT, Claude, Gemini and Grok.
- Officials say the firms deployed multiple accounts and proxy networks to circumvent detection systems and usage restrictions.
- Moonshot’s Kimi K3 model drew significant interest after achieving competitive results against top American AI systems in coding tests.
- Asian technology stocks declined as markets digested implications of intensifying competition between American and Chinese AI developers.
Federal security authorities have leveled accusations against six Chinese AI development firms, including DeepSeek, Alibaba and Moonshot, claiming they leveraged American AI models to enhance their own technology. The NSA, FBI and CISA stated these organizations harvested billions of tokens by submitting millions of queries starting in late 2024.
The agencies said this operation targeted Claude, GPT, Gemini and Grok systems. According to officials, the firms employed proxy infrastructure and numerous accounts to evade detection mechanisms and platform restrictions. American authorities characterized this behavior as malicious knowledge distillation and indicated Chinese government awareness of these operations. Officials noted the campaign launched in late 2024 and persisted across various model platforms through orchestrated account networks and global proxy systems.
Kimi K3 Launch Puts Moonshot in Spotlight
Moonshot captured widespread industry attention following the July debut of Kimi K3, an open-weight system that achieved impressive rankings on a prominent coding evaluation benchmark. The model delivered performance comparable to offerings from Anthropic and OpenAI, prompting renewed examination of Chinese AI advancement velocity.
Kimi K3 demonstrated Moonshot’s capability to produce a competitive system while operating below the investment thresholds of prominent American AI enterprises. CoinDesk reported the launch questioned prevailing beliefs about development expenses and whether advanced AI capabilities would stay concentrated among major U.S. technology firms.
Federal Officials Detail Extensive Extraction Operations
Distillation represents a standard approach in AI model training. Teams submit queries to more capable systems, gather the outputs, and apply that information to refine alternative models. American authorities contend the Chinese enterprises deployed this technique at unprecedented scale.
According to federal agencies, these firms dispatched millions of strategically crafted queries to American AI systems. The objective centered on capturing coding proficiency, mathematical reasoning, and logical capabilities. Officials stated the companies implemented technical measures to bypass safeguards, usage quotas, and monitoring systems across multiple platforms.
Investment Markets Respond to Competitive Pressure
Chinese artificial intelligence equities declined following these revelations amid heightened anxiety about sectoral rivalry. Z.ai shares dropped approximately 27%, while MiniMax experienced roughly 16% losses. Semiconductor stocks also weakened throughout Asian markets as investors reconsidered projections for AI infrastructure investment and model advancement.
The downturn extended to additional markets. Bitcoin alongside other prominent cryptocurrencies declined, while South Korea’s Kospi surrendered another 3.5% during the subsequent trading session. Market participants also questioned whether substantial American expenditures on processors and computing infrastructure would consistently produce superior AI models compared with more economical Chinese competitors.
Get 3 Free Stock Ebooks
Discover top-performing stocks in AI, Crypto, and Technology with expert analysis.
- Top 10 AI Stocks - Leading AI companies
- Top 10 Crypto Stocks - Blockchain leaders
- Top 10 Tech Stocks - Tech giants
