U.S. to Investigate Chinese AI Models Over Potential IP Violations, Treasury Secretary Says

2026-07-21

The BareStory

U.S. Treasury Secretary Scott Bessent announced on Tuesday that the Trump administration will investigate whether Chinese artificial intelligence models have been trained using data taken from American models. Bessent stated that the U.S. government holds the authority to sanction foreign AI models if they are found to be stealing intellectual property from U.S. firms.

Bessent claimed that watermarks belonging to U.S. large language models have been detected on multiple Chinese models, describing the situation as unacceptable. He added that the administration intends to examine the issue in the coming days or weeks. This development comes as Chinese open-weight models, including Moonshot AI’s Kimi K3, increase their competitiveness against American AI products from companies like OpenAI and Anthropic. "Distillation" refers to an AI training method where a smaller model is built using outputs generated by a larger, more powerful model.

To support these concerns, Anthropic sent a letter to the U.S. Senate Committee on Banking, Housing, and Urban Affairs last month, alleging that Chinese tech firm Alibaba carried out the largest known distillation attack against its systems to date. Amid these ongoing technology disputes, the U.S. and China are reportedly planning to hold bilateral talks regarding artificial intelligence in September, with Bessent scheduled to represent the U.S. during the discussions.

While the U.S. government focuses on foreign firms, American AI companies have faced similar intellectual property disputes domestically. Anthropic reached a $1.5 billion settlement in September to resolve a class-action lawsuit with authors who accused the firm of illegally downloading books from pirated databases. In addition, OpenAI and Microsoft faced a 2023 lawsuit from a major newspaper that alleged copyright infringement through the unlicensed use of intellectual property during model training.

Left Perspective

Exposing Systemic IP Hypocrisy
Reforming Domestic Data Extraction
Risking Escalation Over Diplomacy

Right Perspective

Shielding National Strategic Assets
Enforcing Transnational Rule of Law
Deterring Adversarial Market Extraction

Left Perspective

• Exposing Systemic IP Hypocrisy The value of systemic accountability requires holding all powerful institutions to the same legal standards, regardless of their nationality. By focusing aggressively on Chinese distillation attacks, the administration shields domestic tech giants from scrutiny over their own history of intellectual property violations. Specifically, while the government targets foreign firms, American companies like Anthropic recently agreed to a $1.5 billion settlement with authors over pirated databases, and OpenAI continues to face a lawsuit from a major newspaper for unlicensed training.

• Reforming Domestic Data Extraction A reformer framework prioritizes the protection of individual creators, workers, and public goods over the preservation of corporate monopoly profits. This perspective views domestic lawsuits, such as the 2023 copyright action against OpenAI and Microsoft, as evidence of structural exploitation inherent in the current AI development model. Rather than using state power to protect the proprietary data of American tech firms, the policy priority should be restructuring domestic laws to ensure fair compensation for human labor and creative work.

• Risking Escalation Over Diplomacy The reformer lens views diplomatic de-escalation and international cooperation as the most effective paths to long-term global stability. Threatening unilateral sanctions against Chinese AI models ahead of the scheduled September bilateral talks risks creating a hostile, nationalistic tech war that stymies global progress. This adversarial posture undermines the potential for collaborative international guardrails, ultimately increasing geopolitical friction and reducing the likelihood of reaching shared agreements on AI safety.

Right Perspective

• Shielding National Strategic Assets The value of national sovereignty requires the state to defend its borders, economic assets, and critical technological infrastructure from foreign exploitation. Treasury Secretary Scott Bessent's planned investigation into Chinese AI models is a necessary application of state power to protect American innovation. With Chinese open-weight models like Moonshot AI's Kimi K3 rapidly advancing, the government must act to prevent foreign state-backed actors from eroding America's strategic technological edge.

• Enforcing Transnational Rule of Law Preserving established systems of commerce requires the strict enforcement of property rights and the penalization of actors who bypass legal market mechanisms. The discovery of U.S. large language model watermarks on multiple Chinese models represents a clear, unacceptable violation of intellectual property standards that demands a strong sovereign response. Utilizing the authority to sanction foreign models protects the integrity of global market competition and ensures that foreign competitors cannot steal American intellectual property with impunity.

• Deterring Adversarial Market Extraction A realist security perspective views strategic deterrence as the only reliable way to prevent economic attrition and maintain national power. If the U.S. does not respond decisively to massive distillation attacks, such as Alibaba's documented breach of Anthropic, it signals weakness and invites further intellectual property theft. Failing to defend American capital and research risks allowing geopolitical adversaries to free-ride on domestic technological breakthroughs, ultimately threatening long-term national security.

How it may affect me

As a U.S. reader:

• You may face restrictions or lose access to competitive foreign AI products, such as Chinese open-weight models, if the U.S. government exercises its authority to sanction them.

• You could see future changes in how American creators, authors, and journalists are compensated for their work if domestic intellectual property disputes lead to a restructuring of laws surrounding AI training.

• The pace of innovation and the safety of AI technologies you use could be affected if escalating political friction prevents international cooperation and shared global guardrails.

• The market dominance and technological capabilities of domestic AI services you rely on could be preserved if government intervention successfully protects American firms from foreign data extraction.

U.S. to Investigate Chinese AI Models Over Potential IP Violations, Treasury Secretary Says

The BareStory

Left Perspective

Right Perspective

How it may affect me

Read the story at