Zak Kukoff breaks down Trump's AI Action Plan, three executive orders, and the Hill and Valley summit

Summary

Zak Kukoff joined from Washington to break down the Trump administration's AI Action Plan, three executive orders signed at the Hill and Valley Forum, and what the summit revealed about the administration's direction on AI policy.

Hill and Valley

The Hill and Valley Forum, held at the Mellon Auditorium in DC, was organized in roughly 10 days once a slot opened on the president's schedule. What started a couple of years ago as a dinner now operates as a staffed organization of around 30 people. Trump's keynote was the highlight — described as substantive on AI policy and notable for sharp crowd-work targeting Jensen Huang and Doug Burgum. Trump opened by thanking the All-In podcast, pausing pointedly before adding "yes, even Jason" to audible laughter.

The three executive orders

Three AI-related EOs were signed at the event.

• Bias-free AI: Model providers that want to be used in federal procurement must attest to and be transparent about the values embedded in their models. Kukoff frames this as a transparency measure rather than a purely ideological one, even if the political framing centered on removing DEI from AI.
• Data center permitting: Fast-tracks permitting for AI infrastructure buildout, including categorical exclusions from NEPA review — authority the White House can exercise directly via EO.
• Export expansion: Addresses the loosening of chip export controls, consistent with the administration's recent moves on H20 sales.

The AI Action Plan

Beyond the EOs, the broader AI Action Plan functions as a policy wish list — things that require either Congress or state and local governments to act. The most discussed item, a 10-year moratorium on state-level AI regulation, cannot be implemented by EO alone. California is already pushing back on federal preemption claims, and senators like Marcia Blackburn have incentives to protect home-state industries such as Nashville's music business from blanket federal override.

The underlying logic the administration is pushing: in a 50-state patchwork, every AI company defaults to complying with the most restrictive jurisdiction. Trump made that argument directly in his speech.

Copyright and fair use

Kukoff flags the administration's codification of a fair-use standard for AI training data as a meaningful development. The position, consistent with a recent Anthropic court ruling, is that training on legally purchased copyrighted material constitutes fair use — but training on pirated data does not. This is the first time Kukoff says he heard the administration formally adopt that distinction.

Soft power and open source

The AI Action Plan's strong support for open-source and open-weight models is framed by the administration as a national security imperative — getting American models adopted globally as a soft-power lever. That framing creates an opening for Meta's Llama project, though Kukoff notes uncertainty about whether Llama 5 will be open source and whether Zuckerberg has fully landed with this administration the way Jensen Huang has. Jensen's success with the White House is partly attributed to interpersonal skill and partly to running the most valuable company in the world. Zuckerberg made a visible effort to engage but the relationship hasn't solidified in the same way.

With OpenAI launching an open-source model, the title of "America's national open-source champion" is now genuinely contested between Meta and OpenAI — and Kukoff suggests that title carries real political value even if its direct commercial value is unclear.

The talent gap no one addressed

One conspicuous omission from the AI Action Plan: a targeted talent recruitment strategy. Around 75% of Meta's superintelligence team reportedly hold O-1 visas or are non-US citizens. The DeepSeek and Manus teams signal what concentrated foreign AI talent can produce. Kukoff says people inside the administration are privately thinking about an "Operation Paperclip" equivalent — a targeted program to bring the world's top AI researchers to the US — but it was not part of the formal AI Action Plan and immigration politics make it difficult to address publicly. The argument is that this cohort, perhaps 50 people globally, is a categorically different policy question from border enforcement.