Officials from the Cyberspace Administration of China (CAC) visited the offices of DeepSeek and Moonshot AI to interrogate executives and staff. The state cyber regulator officially launched an investigation into allegations that both artificial intelligence labs covertly funneled user data to Anthropic’s servers in the United States.
The regulator’s primary focus centers on the nature of information crossing borders. The CAC is currently reviewing data flows to determine whether sensitive information from law enforcement, the military, or state-affiliated corporate entities reached US servers through DeepSeek and Moonshot’s activities.
The probe stems from a 154-page document. On September 10, Anthropic published its fourth threat intelligence report accusing seven Chinese AI labs of illegal distillation practices via networks of fraudulent accounts. The seven entities named in the report include major players such as Alibaba, Moonshot, DeepSeek, Zhipu, MiniMax, SenseTime, and Xiaomi.
Model distillation is fundamentally the process of training a smaller AI model using outputs from a larger one. Anthropic’s objections arose because the training process was carried out using fake accounts to scrape data rather than through legitimate licensing channels.
Narrowed Down to Two Primary Targets
The CAC initially summoned all seven companies listed in Anthropic’s report for questioning. As interrogations progressed, the regulator’s focus narrowed sharply. Officials have now turned their full attention solely toward DeepSeek and Moonshot, though no official penalties have been announced for either company.
The probe comes at an awkward time for the companies’ diplomatic engagements. Globally, DeepSeek is scheduled to present on artificial intelligence risks before the UN Security Council this week. That schedule places DeepSeek representatives on the same stage as Anthropic CEO Dario Amodei.
The UN presentation also takes place just one day before a summit between Donald Trump and Xi Jinping on September 24. Artificial intelligence is confirmed to be on the list of topics discussed by the two world leaders.
Threat to a $3 Billion IPO
For Moonshot, regulatory questioning comes right in the middle of their confidential initial public offering (IPO) process on the Hong Kong stock exchange. Filing documents were submitted on September 3, with the company aiming to raise $3 billion at a projected $50 billion valuation.
Intervention from the cyber watchdog threatens to disrupt the company’s path to public markets.
The domestic clampdown presents its own irony. China is now using Anthropic’s leak detection report as a blueprint to rein in its domestic AI labs. An intelligence document intended to expose external IP scraping is instead being wielded by regulators to uncover domestic operational practices. Reported by Decrypt.
Read also: Following $107,539 White House Teleprompter Scandal, CFTC Targets Speech Bets on Polymarket
Disclaimer: This article is for informational and educational purposes only, not financial advice. Cryptocurrency assets are highly volatile and carry significant risk. Always do your own research (DYOR) and never invest more than you can afford to lose.
