Regulation•

David Friedberg Calls a Chinese AI Models Disclosure Plan the Path of Tyranny

On the November 20 All-In episode, David Friedberg attacked a Trump administration idea to make US companies disclose their use of Chinese AI models, warning of an Orwellian outcome.

Written By :MINRK
David Friedberg Calls a Chinese AI Models Disclosure Plan the Path of Tyranny

Photo by Nate Grant on Unsplash

1. Friedberg Takes Aim at a Proposed Disclosure Rule

All-In Podcast co-host David Friedberg used the show's November 20, 2026 episode to condemn a Trump administration proposal that would require American companies to publicly disclose when they use Chinese AI models. He argued that any move to curb, restrict or label open-source AI would limit human prosperity and individual liberty, calling the approach “the path of tyranny.”

The remarks came two days after the proposal surfaced publicly on November 18. The idea under consideration would not ban Chinese models or impose heavy restrictions on them. It would instead make their use a matter of public record, so that customers, regulators and investors could see which US businesses rely on technology developed in China.

For Friedberg, that distinction offers little comfort. In his view, requiring companies to identify the models they run is the opening move in a broader campaign against openly available software, and the costs of that campaign would fall on ordinary users rather than on the governments and corporations that might benefit from it.

2. What Does the Chinese AI Models Proposal Actually Require?

The proposal would compel US companies to state publicly that they use Chinese AI models. It stops short of prohibiting those models, and it does not, as described, place technical limits on how they can be deployed. Disclosure, rather than restriction, is the core mechanism.

Important details remain unsettled. No formal rulemaking timeline has been confirmed, and the enforcement mechanics have not been spelled out. It is not yet clear which agency would administer the requirement, what penalties would apply for failing to comply, or how the government would define a Chinese AI model in the first place. For now, the measure is a proposal under discussion, not a binding rule.

Disclosure regimes are a familiar policy tool in Washington. Rather than outlawing a practice, they force it into the open and let market pressure, public opinion and political scrutiny do the rest. Supporters of that approach tend to see it as a lighter touch than an outright ban. Critics argue that mandatory labeling can carry a stigma that works almost like a prohibition, pushing companies away from a technology without any formal finding that it is unsafe.

3. The Case Friedberg Made on Air

Friedberg framed his objection around two values he said open-source AI protects: prosperity and liberty. Efforts to restrict or label openly available models, he argued, would limit both. He warned that continuing down that road could create an “Orwellian state,” and he said open-source AI should be encouraged rather than discouraged.

His reasoning centered on who ends up in control when open alternatives disappear. If openly published models are pushed aside, the groups left standing would be either tyrannical actors or regulated monopolies facing little real competition. In that world, a small number of closed-source providers aligned with government would hold power over what people can do with the technology.

That argument turns the usual national security logic around. Proponents of disclosure see scrutiny of Chinese models as a way to protect American companies and their customers. Friedberg sees scrutiny of open code as a gift to closed, well-connected incumbents that publish nothing about their own systems and therefore never face the same questions.

4. Why Competition Sits at the Center of the Argument

Friedberg's warning about regulated monopolies draws on a long-running theme in technology policy. When a market has only a few approved suppliers, those suppliers face less pressure to cut prices, improve quality or open up their products. Rules that raise the cost of using alternatives, even indirectly, can entrench whoever is already on the inside.

In the AI market, openly available models act as a ceiling on what closed providers can charge and a floor on what users can do without permission. A company that can download a capable open model has an outside option if a proprietary vendor raises prices or changes its terms. Remove or stigmatize that option, the argument goes, and the bargaining power shifts toward the largest closed labs.

That is why Friedberg treated a disclosure requirement, which on its face restricts nothing, as a serious matter. His concern is less about the paperwork itself than about the direction of travel: a policy that singles out certain openly published models today could, in his telling, become a precedent for constraining open software more generally, with competition as the casualty.

The All-In Podcast is widely followed in technology and investment circles, which gives the hosts' views on AI policy an audience well beyond casual listeners.

5. Apache and Firefox as Historical Precedent

To support his point, Friedberg reached back to the early internet. He cited the Apache web server software and the Firefox browser as open-source projects that helped keep core internet technology accessible to everyone.

Apache, developed under the Apache Software Foundation, became one of the most widely used pieces of software for running websites, giving businesses and individuals a free alternative to proprietary server products. Firefox, from the Mozilla Foundation, offered an open browser at a time when a handful of large companies dominated how people reached the web. Friedberg presented both as examples of openness acting as a check on any single company's ability to control a foundational technology.

The comparison implies that AI models are becoming similar building blocks: infrastructure that many products sit on top of. If that is true, the question of who can freely use, inspect and modify those building blocks could shape the competitive landscape for years, much as open web software did two decades ago.

6. Why Chinese AI Models Draw Scrutiny in Washington

The disclosure idea did not emerge in a vacuum. Lawmakers have already been asking individual companies about their use of Chinese AI. The House Select Committee on China has sent inquiries to companies including Airbnb and DoorDash about the Chinese models they rely on.

Airbnb has featured prominently in that discussion as a user of Qwen, a family of AI models developed by Alibaba. Chinese developers have released a number of capable models with openly available weights, meaning companies can download them, run them on their own systems and adapt them to their own needs. That flexibility, combined with low costs, has made such models attractive to businesses looking to build AI features without paying for access to proprietary American systems.

A disclosure rule would change the character of that oversight. Instead of lawmakers questioning firms one at a time, every company using a Chinese model would have to answer the same question publicly. What is currently a targeted congressional inquiry would become a standing obligation across the corporate landscape.

7. Open Weights and the Wider Open-Source Debate

Much of the tension stems from how open-source AI works. When a developer publishes a model's weights, anyone can run it, study it or build on it. That openness is exactly what Friedberg praised, but it also means a model's origin can travel far beyond its creator, embedded in products that end users may never associate with the original developer.

For policymakers worried about national security, that diffusion is the problem. A model trained in China could sit underneath a customer service tool, a coding assistant or an internal business application in the United States without most people knowing. Disclosure is meant to make that dependency visible.

For open-source advocates, the same diffusion is the point. Openly available models lower the cost of building with AI, spread capability beyond a few dominant labs and make it harder for any single gatekeeper to decide what can be built. Friedberg's warning reflects a concern that targeting models by country of origin could become a template for scrutinizing openly published software more broadly.

8. Anthropic and Nvidia Also on the Agenda

The open-source segment was one part of a wide-ranging episode. The same November 20 broadcast also covered Anthropic's IPO and the stock market difficulties facing Nvidia, two of the most closely watched stories in technology and markets this month.

The pairing underlines how central AI has become to both policy and investing conversations. Decisions about which models companies can use, and on what terms, feed directly into questions about which AI providers will capture business and how investors value them.

9. What Companies and Investors Should Watch Next

In practical terms, the immediate impact is limited because no rule exists yet. What exists is a reported proposal, a set of congressional inquiries and a sharp public disagreement over the right response. Companies are not currently obligated to disclose anything under the plan.

Several developments could change that picture. The first is how the companies already named in congressional inquiries, such as Airbnb and DoorDash, respond in public. Their answers could become a template for how other businesses explain their use of Chinese models, whether or not a formal rule follows.

The second is the definition at the heart of any rule. A narrow definition covering only models released directly by Chinese developers would affect a limited set of companies. A broader one that reaches fine-tuned derivatives, or models built on top of Chinese base weights, would touch far more businesses, because so much AI development now involves adapting existing open models rather than training new ones from scratch.

Friedberg's position and the administration's emerging approach rest on opposite bets. He is wagering that openness is the better long-term defense, keeping power distributed and competition alive. The proposal under discussion in Washington puts visibility first, starting from the premise that the government and the public should know which companies depend on Chinese AI models before deciding what, if anything, to do about it.

Related Articles

NEWSLETTERS

Don't miss another story.

Subscribe to the MINRK Newsletter today.

Tick the newsletters you want. No selection? You'll get MINRK Headlines.

By signing up, you will receive emails about MINRK products and you agree to our terms of use and privacy policy.

MINRK Daybook

Market analysis for crypto traders and investors.

EVERY WEEKDAY

MINRK for Advisors

Defining crypto, digital assets and the future of finance for financial advisors.

EVERY THURSDAY

MINRK Protocol

Exploring the tech behind crypto one block at a time.

WEEKLY

MINRK Long & Short

A must read for institutions. Insights, news and analysis delivered weekly.

EVERY WEDNESDAY

MINRK Headlines

The biggest crypto news and ideas of the day.

EVERY WEEKDAY

MINRK Policy

Examining the intersection of cryptocurrency and government.

WEEKLY

MINRK Research

Join thousands of readers who rely on MINRK for data-driven insights on the latest digital asset trends.

MONTHLY