China’s decades of cyber theft may finally pay off in the age of AI

China’s decades of cyber theft may finally pay off in the age of artificial intelligence. China has been stealing American data for years, from trade secrets to manufacturing know-how to sensitive research. However, very few of these assets have reached the black market, and China does not openly use much of the data it actively acquires. That’s because effectively appropriating so many stolen secrets is a daunting task and is severely limited by personnel capabilities, language skills, and personal expertise.

AI will change that equation and have a profound impact on America’s economic and security interests.

China is now leveraging its unique data stockpile to train models that discover patterns humans will never see, identify new opportunities, and create new products across industry, science, and defense. I can even do it. Powerful “training data” makes AI-driven insights more robust and actionable. But few are wary of how AI can exponentially increase the value of hacked corporate data to strategic competitors, especially China. reportedly It outperforms the world in 37 out of 44 technology areas.

Last year, just weeks before ChatGPT showed the world what could be done with generated AI, the Biden administration national security strategy It affirmed the oft-quoted adage. Economic security is national security. Emerging AI tools now offer the Chinese Communist Party an unprecedented opportunity to derive economic benefits from years of accumulated secrets, posing significant risks to U.S. national security interests.

The US is not sharing China’s windfall profits. U.S. trade secrets are necessarily private and siled across thousands of private companies. No single U.S.-based company can hold all company data, much less feed that data into AI models to identify new market opportunities. Simply put, in addition to the vast amount of domestic data it collects on its citizens and businesses, China has built a vast collection of AI training data from countless non-public sources.

This is why China’s longstanding hacking operations are potentially so damaging. We must assume that China can leverage its treasure trove of stolen assets and domestic data to accelerate the development of new products, gain a competitive edge, and identify critical vulnerabilities in the American market. . They have both the means and the motivation to seize this opportunity. And comparable capabilities are far less likely to be available in non-authoritarian states where sensitive corporate data is properly distributed across private entities.

Moreover, China’s AI-derived products and services will be difficult to detect as spin-offs are formed by so many data points. American businesses, the government, and Congress must prepare for a future of unfair competition in which Chinese products are built on illegal caches of intellectual property from around the world.

Countering this threat requires strengthening deterrence and improving cyber resilience.Law enforcement agencies in the United States and Australia recently called out China supported data collection operations and encouraged private companies to secure critical systems. That’s all well and good, but a more carefully coordinated and aggressive campaign is needed to throw China off course. Meaningful progress in securing U.S. interests requires that, at a minimum, policymakers recognize how generative AI can increase the value of stolen data and treat this economic challenge as a national security threat. be.

Jeff Greene is senior director of the cybersecurity program at the Aspen Institute. He previously served as director of cyber response and policy in the National Security Council’s Cyber ​​Directorate. Josh Lawson advises the Aspen Institute on artificial intelligence and democracy. Previously, he served as Head of Elections and Emerging Risks at Meta Platform, Inc.