'China already has spies' at US AI labs, claims ex-Anthropic researcher
- Former OpenAI and Anthropic researcher Jacob Coxon claims a 90% certainty of Chinese espionage within top US AI labs.
- Coxon warns that any American lead in AI development is effectively neutralized by Beijing's ability to steal proprietary progress.
- Chinese nationals currently constitute 38% of top-tier AI talent in the US, complicating security efforts.
- Treasury Secretary Scott Bessent dismisses the threat as 'distillation,' comparing it to students copying homework.
Brief Summary
A former researcher from the two most prominent AI labs in the country has dropped a bombshell claim, asserting that Chinese intelligence has likely already breached the inner sanctums of OpenAI and Anthropic. Jacob Coxon, who left the industry citing safety concerns, argues that the race for AI dominance is a farce if the finish line is being handed to Beijing via corporate espionage. While critics call his 90% certainty claim 'irresponsible,' the reality remains that the US relies heavily on a workforce where Chinese nationals now outnumber Americans in top-tier roles.
Why This Matters
If these labs are indeed compromised, the billions of dollars in private and public investment poured into American AI are essentially subsidizing China's own technological advancement. For you, this means the 'AI revolution' may not result in a uniquely American competitive edge, but rather a global race where the fruits of your digital labor and data are being harvested by foreign adversaries. This security failure could accelerate the timeline for advanced AI tools to be used against American interests, affecting everything from cybersecurity and financial stability to the integrity of the information you consume online.