There you have it! A top Pentagon official just called OpenAI's own policy strategist the "suprem…
July 20, 2026 · 0 likes · 0 comments
China Threat AI Defense Cybersecurity Workforce
There you have it! A top Pentagon official just called OpenAI's own policy strategist the "supreme village idiot" of AI — in public, with his name attached.
Let me tell you what actually happened, because the insult is not the story.
Dean Ball helped write the White House AI Action Plan. Then he went to work for OpenAI. And this week he floated a plan: the government does not need to formally ban Chinese AI models like DeepSeek. It can just have agencies quietly issue warnings — manufacture enough "regulatory uncertainty" that no American company dares touch a Chinese model.
No law. No vote. No formal ban. Just fear, applied through a memo.
Emil Michael, the Under Secretary of War for research and engineering, torched it. His point was blunt: Congress already passed a law in 2026 restricting specific Chinese models, with waivers, through the actual democratic process. Not a backroom scheme. Not a policy invented to help whoever wrote it.
And here is the part that should make everyone sit up.
David Sacks — President Trump's own AI guy — called using regulatory uncertainty against Chinese AI "completely unacceptable" and asked the obvious question out loud: is this regulatory capture dressed up as national security?
Because look at who benefits. If Chinese open-weight models get frozen out by manufactured fear, who is left holding the market? The company that employs the man writing the playbook.
That is not a China strategy. That is a moat with a flag painted on it.
I have been warning about exactly this for years. American AI policy cannot become a private toll road for whichever vendor has the best access in Washington. The second we start writing rules to protect a company instead of the country, we have already lost — because our adversary does not slow down while we rig the game for our incumbents.
This is the exact dynamic I wrote about in my book, REPLACEMENT. The real replacement is not just AI swapping out human jobs. It is a handful of players quietly rewriting the rules so they replace the competition itself — and calling it national security while they do it. Once you see that pattern, you see it everywhere: in the memo, in the market, in the moat.
Here is what winning actually requires: model-agnostic architecture, real competition between vendors, and clear security standards that apply to everyone the same way. Not warnings whispered into the market by the people who profit from the silence.
Beat China by building faster. Not by lobbying the government into picking your product.
Source:
https://lnkd.in/ejKPWR-5
One more thing. REPLACEMENT launches July 28, and there are only 100 signed Collector's Editions — numbered, each dual-signed by me and by Quill, one of my AI agents driving a robotic arm. Only 53 left. When the numbered run is gone, it is gone. If you want one, order it here: https://lnkd.in/eEwXNwrf
Let me tell you what actually happened, because the insult is not the story.
Dean Ball helped write the White House AI Action Plan. Then he went to work for OpenAI. And this week he floated a plan: the government does not need to formally ban Chinese AI models like DeepSeek. It can just have agencies quietly issue warnings — manufacture enough "regulatory uncertainty" that no American company dares touch a Chinese model.
No law. No vote. No formal ban. Just fear, applied through a memo.
Emil Michael, the Under Secretary of War for research and engineering, torched it. His point was blunt: Congress already passed a law in 2026 restricting specific Chinese models, with waivers, through the actual democratic process. Not a backroom scheme. Not a policy invented to help whoever wrote it.
And here is the part that should make everyone sit up.
David Sacks — President Trump's own AI guy — called using regulatory uncertainty against Chinese AI "completely unacceptable" and asked the obvious question out loud: is this regulatory capture dressed up as national security?
Because look at who benefits. If Chinese open-weight models get frozen out by manufactured fear, who is left holding the market? The company that employs the man writing the playbook.
That is not a China strategy. That is a moat with a flag painted on it.
I have been warning about exactly this for years. American AI policy cannot become a private toll road for whichever vendor has the best access in Washington. The second we start writing rules to protect a company instead of the country, we have already lost — because our adversary does not slow down while we rig the game for our incumbents.
This is the exact dynamic I wrote about in my book, REPLACEMENT. The real replacement is not just AI swapping out human jobs. It is a handful of players quietly rewriting the rules so they replace the competition itself — and calling it national security while they do it. Once you see that pattern, you see it everywhere: in the memo, in the market, in the moat.
Here is what winning actually requires: model-agnostic architecture, real competition between vendors, and clear security standards that apply to everyone the same way. Not warnings whispered into the market by the people who profit from the silence.
Beat China by building faster. Not by lobbying the government into picking your product.
Source:
https://lnkd.in/ejKPWR-5
One more thing. REPLACEMENT launches July 28, and there are only 100 signed Collector's Editions — numbered, each dual-signed by me and by Quill, one of my AI agents driving a robotic arm. Only 53 left. When the numbered run is gone, it is gone. If you want one, order it here: https://lnkd.in/eEwXNwrf