Meet the Seven People Deciding America's AI Policy Right Now
A senior White House official told Wired it is 'an argument with ten sides'. Here is who is pulling the strings on US AI rules, what each one wants, and what it means for the rest of us.
Key points
- A senior White House official described the internal debate over US AI policy as "an argument with ten sides", according to Wired.
- Commerce Secretary Howard Lutnick is exploring incentives for US labs to build open-weight models to counter China; Treasury Secretary Scott Bessent is threatening sanctions against Chinese AI labs.
- National Cyber Director Sean Cairncross helped draft President Trump's June 2 executive order setting a framework for assessing the most powerful AI models.
- Former AI czar David Sacks, who left his official role in March 2025, is still actively shaping policy through his X account, which has 1.6 million followers.
- White House Chief of Staff Susie Wiles has the final say on which proposals reach the president.
American AI policy isn't being shaped in one room by one team. It's being pulled in multiple directions by a loose group of officials spread across several departments, each with a different idea of what the US should do about China's fast-improving AI.
Wired reported this week on the full cast. Here's what you actually need to know.
Who holds the most power?
Susie Wiles, Trump's chief of staff, holds the final filter. Every proposal lands on her desk before it reaches the Oval Office, and her decision is treated as final.
Below her, two cabinet members are competing for influence.
Commerce Secretary Howard Lutnick controls export controls, the rules deciding which US technology can be sold abroad. He's been exploring ways to incentivise American AI labs to build their own open-weight models (AI systems whose underlying code is made public, so anyone can download and build on them) to compete with Chinese ones. He recently played down the threat from Moonshot AI's Kimi K3 model after his team found it scored worse than leading US models on standard performance tests. We covered the Kimi K3 reaction in Washington as far back as 18 July.
Treasury Secretary Scott Bessent is taking a harder line. He publicly called distillation "IP theft" last week. Distillation is when a company trains a cheaper AI model by having it learn from a more powerful one, sometimes without permission. Bessent has threatened Chinese labs with sanctions or placement on the Entity List, a government blacklist that cuts companies off from American suppliers.
What are they actually fighting about?
The core disagreement is whether the US should clamp down on Chinese AI, risking a trade confrontation, or trust American companies to outrun China on their own.
Former AI czar David Sacks, who still talks directly to the president even after leaving his White House role in March, argues the US is already ahead. On 23 July 2026, he told his 1.6 million X followers that "as long as we keep releasing, we will stay ahead".
National Cyber Director Sean Cairncross disagrees. A former campaign lawyer with no tech background, he's focused on national security risk and pushed the language in the June 2 executive order that sets out how the government will assess the most powerful AI models.
| Official | Department | Position on Chinese AI |
|---|---|---|
| Susie Wiles | White House | Final gatekeeper, no public stance |
| Howard Lutnick | Commerce | Middle ground, open to incentives |
| Scott Bessent | Treasury | Most aggressive; threatens sanctions |
| Sean Cairncross | National Security | Hardline on security risks |
| David Sacks | Ex-White House | Hands-off; trust US labs to win |
What does this mean for ordinary people?
For now, it mostly means uncertainty. Businesses using AI tools built on Chinese models could face new rules with little warning if Bessent's harder line wins out. Workers wondering whether to bet on AI skills should know that Washington itself hasn't decided which way to pull.
Watch what Susie Wiles approves, not what any individual official posts on social media. When a proposal clears her desk, that's when the rules actually change.
Common questions
What is an open-weight AI model and why does it matter?
An open-weight AI model is one where the creator publishes the underlying instructions so anyone can download and build on it for free. That makes it very hard to restrict, which is why Chinese open-weight models worry US officials even when export controls block the sale of American chips.
Could new AI rules affect me or my business?
Possibly, if you use software built on Chinese AI models or sell technology internationally. Sanctions or Entity List additions could cut off access to certain tools. For most consumers using mainstream apps the near-term impact is small, but it's worth checking where your AI tools originate.
