AI models iterate every six months, and White House Office of Science and Technology Policy director Kratsios made it clear this week at a YC open class: if regulation can't keep this pace, reality will leave it behind. The former Scale AI chief operating officer now runs U.S. national strategy on AI and emerging technology.
What this is
Kratsios's talk centered on four points. One: the White House explicitly backs open-source AI, treating it as the key to U.S. technological leadership rather than something to restrict. Two: regulation design must avoid one trap—letting large companies use compliance costs to squeeze out smaller ones, creating new moats. Three: some widely discussed AI risks are exaggerated; what should actually be regulated is identifiable, concrete harm, not vague notions of "AI being too powerful." Four: Washington wants small companies and founders in the policy room, not just the OpenAIs, Anthropics, and Googles of the world.
Industry view
Supporters argue the open-source path keeps the U.S. AI ecosystem vibrant and leaves room for enterprises to build their own models rather than getting locked in by a few vendors. The "anti-moat" framing on regulation is also seen by some legal scholars as a continuation of the 1990s Microsoft antitrust line of thinking.
Criticism is just as sharp. AI safety researchers broadly worry that open-sourcing frontier models means Chinese labs can replicate U.S. advances at zero cost, and that the White House is effectively giving away the technological edge for free. Others note that "adjusting regulation every six months" is nearly impossible under congressional timelines—an average bill takes 1–3 years to pass, during which AI models will have iterated multiple generations. Some also point out that Kratsios himself comes from Scale AI (an AI data infrastructure company valued at tens of billions of dollars), so his "small companies" may not be what most people would actually call small.
Impact on regular people
For enterprise IT: With open-source model weights continuing to open up, the cost of building proprietary AI tools on open foundations and compliance costs are both falling—this is a direction worth probing in IT procurement.
For personal careers: Hybrid talent who understands both AI and policy is becoming scarce. Over the next 2–3 years, these people will command more value at consultancies, government relations roles, and corporate strategy departments.
For the consumer market: Open-source models may accelerate the entry of AI assistants and AI search into consumer apps. Regular users will get cheap or even free AI features faster, but privacy boundaries will become blurrier.