Humanity CEO Dario Amodei worries that a spy from China, who is likely to get expensive “algorithm secrets” from top American AI companies.
Speaking at the Council on Diplomatic Relations on Monday, Amodei said China is known for its “large industrial spy” and that AI companies like humanity are almost certainly targeted.
“Many of the secrets of these algorithms have a $100 million secret, a few lines of code,” he said. “And you know, I'm sure there are people trying to steal them, and they may be successful.”
More help from the US government in defending this risk is “very important,” Amodei added, but does not specify exactly what kind of assistance is needed.
Humanity declined to comment on TechCrunch in particular about his remarks, but earlier this month he mentioned human recommendations for the White House's Science and Technology Policy (OSTP).
In its submission, humanity argues that the federal government needs to partner with AI industry leaders to enhance security at the Frontier AI Lab, including working with the US intelligence agency and its allies.
This statement follows Amody's more important stance on China's AI development. Amodei calls for strong US export controls for AI chips to China, and Deepseek says it has achieved the “worst” in critical biological weapon data safety tests carried out by humanity.
Amodei's concerns are central to China, which uses AI for authoritarian and military purposes, as he laid out the essay “The Machine of Loving Grace” and elsewhere.
This kind of attitude should not be collaborated with less in AI to avoid competition from countries that build systems that are strong enough for humans to control.