White House SIF Co-Chair: "We Will Control AI Companies That Do Not Cooperate with Government"

▲ AI (File Photo)

Emil Michael, U.S. Department of Defense Under Secretary and co-chair of the White House Superintelligence Force (SIF), stated that if artificial intelligence (AI) companies do not cooperate with the government, various administrative measures such as export controls will be mobilized to control their actions.

Michael is the official who pushed for designating Anthropic as a "supply chain risk company" after it refused to remove safety guardrails.

The tech-focused media outlet The Information reported on the 19th (local time) that Michael made these remarks in an interview with the publication.

Michael argued that it is the responsibility of the Trump administration to control AI companies such as Anthropic.

Thus far, the Trump administration's approach to AI oversight has largely focused on voluntary measures, such as a non-binding joint agreement signed by major AI companies following a White House summit late last month.

However, Michael argued that it is appropriate to apply stricter check-and-balance measures to AI companies if necessary.

He cited the export controls from last June that blocked users outside the United States from accessing Anthropic's AI model as an example.

"We have a lot of tools at our disposal, ranging from restricting the export of technology deemed capable of benefiting adversaries, to full-blown export controls, to restricting foreign nationals from using the software," he said. "You can trust that if a real problem arises, the administration and the president will respond urgently."

He also objected to the idea of anthropomorphizing AI models, citing Anthropic's policy allowing its model Claude to end conversations with malicious users as an example.

"If we don't want humans to lose control, we must not give models the authority to control humans," he said.

He added that as long as "such moral principles are embedded in the software," he would not allow Anthropic to be used within the Department of Defense.

Michael criticized cutting-edge AI companies that call for stronger regulations for safety as acting like drunk drivers who want someone else to stop the car instead of stopping it themselves.

He also acknowledged that national security risks related to AI are growing.

"It is a bit concerning that China's open-source models are gaining a dominant position in the U.S., because those companies are controlled by the Chinese government," he said.

While stating that "one way to counter the growing use of these models in the U.S. is to audit model weights," he added that he prefers addressing this issue through the growth of American open-source alternatives such as Reflection, Thinking Machines, and NVIDIA's NeMoTron.

However, he also acknowledged that none of these companies have yet reached the cutting-edge level.

Michael said corporate espionage is also a concern.

Referring to a recent case where an OpenAI employee of Saudi nationality installed a wiretapping device, he noted, "Not all foreign nationals are bad, but there is a certain amount of risk associated with foreign nationals."

He argued that rather than increasing government controls further, companies need to heighten their vigilance.

※