Howard Lutnick wants to beat Chinese AI by opening up American models, not locking them down

The default assumption in Washington’s AI policy conversation has been that American advantage is best preserved through restriction: export controls, access limits, and keeping frontier model weights out of public reach. Howard Lutnick, according to Nathaniel Whittemore, does not share that default. The Commerce Secretary would prefer to combat Chinese AI with competition. He has pitched incentives for US labs to open source their models and has spoken with multiple labs about the idea.

That framing, if it reflects genuine departmental intent, puts Lutnick at odds with the dominant tendency in federal AI policy. The Biden-era executive orders on AI and the subsequent chip export controls both operated from a containment logic: limit what adversaries can reach, slow their progress by restricting access to tooling and talent. Lutnick’s reported position inverts that logic. Instead of asking what Beijing cannot have, it asks what American labs should be encouraged to share, and what the federal government is willing to offer in exchange.

The mechanics of such an incentive scheme are not specified in Whittemore’s account, and that gap matters. Incentives for open sourcing could mean tax treatment, favorable regulatory posture, preferential access to federal compute infrastructure, or something else entirely. The form of the inducement shapes who it benefits and how labs would respond. A cash subsidy reaches labs regardless of their current business model. Regulatory relief is more valuable to a company facing near-term compliance costs. Compute access disproportionately benefits smaller or newer entrants. Without knowing which instrument Lutnick has floated in those conversations, the proposal remains a direction rather than a policy.

Commerce Secretary Howard Lutnik would prefer to combat Chinese AI with competition. He has pitched incentives for US labs to open source their models and spoken with multiple labs about the idea. Nathaniel Whittemore

What the claim does establish is that those conversations are happening. Lutnick has spoken with multiple labs about the idea, Whittemore reports. That is not a think-tank white paper or a congressional staffer’s trial balloon. It is the Commerce Secretary having direct conversations with the institutions whose behavior any such policy would need to change. Whether those conversations have produced alignment, skepticism, or something in between is not part of the account. But the fact that they are occurring at all signals that open sourcing as a strategic tool, rather than a default or a concession, is being seriously considered at a senior level inside the administration.

The competitive logic behind the position has some purchase. If the concern is that Chinese open-source models, among them DeepSeek, give Beijing’s ecosystem a durable foundation to build from, then matching that openness with American-origin models changes the terrain. Developers outside China, given a credible, well-resourced American open-source alternative, have less reason to anchor their work to a Chinese-origin stack. The argument is not that open sourcing eliminates risk. It is that the risk of ceding the open ecosystem to Chinese models is larger than the risk of releasing American weights.

That argument is contested. Critics of open sourcing frontier models argue that weights, once released, cannot be recalled, that adversaries benefit from open releases as readily as allies, and that the competitive advantage of American labs is partly a function of the moat their closed models represent. Lutnick’s pitch, as Whittemore describes it, does not engage those objections directly, because Whittemore is reporting the Secretary’s orientation, not the full internal policy debate. What the claim captures is a posture: the top trade and commerce official in the United States government thinks competition is the right frame, not containment.

That posture, if it hardens into policy, would represent a significant shift in how the federal government positions itself relative to the open-source question. Labs that have stayed closed partly out of regulatory caution might find the calculus different if Washington is offering something in return for openness. Labs already committed to open releases would find their position newly aligned with federal priorities. The shape of the American AI ecosystem, and by extension the shape of the global one, does not turn on a single official’s preference. But a Commerce Secretary with a direct line to lab leadership and a concrete proposal in hand is not nothing. The conversation is further along than the public record has reflected.