Alexandra Spills the Tea on China's AI Heist: Tens of Thousands of Fake Accounts Caught Stealing US Secrets

This is your Cyber Sentinel: Beijing Watch podcast. Good morning, I'm Alexandra Reeves, and this is Cyber Sentinel: Beijing Watch. Let's dive into what's happening in the cyber threat landscape right now. The White House just dropped a significant memo through Michael Kratsios, director of the Office of Science and Technology Policy, revealing that Chinese entities are running industrial-scale campaigns to steal American AI intellectual property. We're talking tens of thousands of proxy accounts, jailbreaking techniques, and coordinated extraction of capabilities from frontier AI systems. This isn't amateur hour anymore, listeners. What makes this particularly alarming is the infrastructure behind it. According to reporting from the Dutch military intelligence agency MIVD, China's cyber-espionage capabilities now match the sophistication of the United States. Dutch Vice Admiral Peter Reesink stated that these operations are extremely capable and organized in complex ways, with Beijing primarily targeting Western defense industries and arms producers to gain access to military technologies and identify vulnerabilities. The technical sophistication extends to military applications. A procurement notice from a military unit in Anhui Province reveals the People's Liberation Army is integrating DeepSeek, a Chinese AI model, into a cybersecurity system designed for intelligent attacks and penetration testing. Another RFP specifically calls for DeepSeek deployment in psychological attack and propaganda systems. These aren't theoretical threats, listeners. They're operational requirements being documented in procurement channels. The compute foundation matters here. Many of these Chinese models, including DeepSeek, were trained on US-designed chips, creating a direct pipeline between American computing power and Chinese military capabilities. Some were reportedly distilled from American models themselves, meaning our own innovation is being weaponized against us. On the defensive side, NASA's Office of Inspector General documented a multi-year spear-phishing campaign where a Chinese national posed as US researchers, targeting NASA employees and defense-related software systems. This represents the human element of cyber operations that technical defenses alone cannot stop. From an international response perspective, the US State Department has directed diplomats worldwide to flag these risks to allied nations. The timing matters too, given that these revelations emerge ahead of a scheduled summit between US and Chinese leaders next month. The geopolitical tension is real. For security measures, organizations need to implement strict access controls on AI systems, monitor for unusual proxy account behavior, deploy advanced email authentication protocols, and conduct regular security awareness training focused on social engineering tactics. At the strategic level, policymakers need to reassess technology export controls and compute availability to This content was created in partnership and with the help of Artificial Intelligence AI.