US Backs Down on DeepSeek CXMT Blacklist: Washington 2026

Key Points

• The United States has postponed adding Chinese AI startup DeepSeek, memory chip manufacturer CXMT (ChangXin Memory Technologies), and more than 100 other companies to its trade blacklist known as the Commerce Department’s Entity List
• Two sources familiar with the matter confirmed the Trump administration backed down from the move to avoid escalating tensions with Beijing
• The vast majority of the 100+ companies on the approved list are Chinese, specializing in advanced semiconductors and AI development
• A joint interagency committee had approved adding DeepSeek, CXMT, and other companies to the Entity List last year
• A senior U.S. State Department official told Reuters in June 2025 that DeepSeek had supported Chinese military and intelligence operations
• The State Department official added that DeepSeek attempted to use shell companies in Southeast Asia to illegally access advanced U.S. chips
• Anthropic announced in February 2026 that it detected a hijacking campaign by DeepSeek and two other Chinese AI labs (Moonshot and MiniMax) targeting its Cloud AI platform
• The hijacking campaign involved over 16 million fraudulent API exchanges using approximately 24,000 fake accounts to steal Claude AI capabilities through “distillation” attacks
• DeepSeek was responsible for 150,000 exchanges, Moonshot for 3.4 million, and MiniMax for 13 million of the fraudulent exchanges
• The Entity List has not received new entries since October 2025, marking the longest interval without updates in over ten years
• OpenAI also warned lawmakers that DeepSeek was targeting its AI models
• CXMT was previously designated as a Chinese military company by the Defense Department under the Biden administration

Washington (Evening Washington News) June 16, 2026 – The United States has officially backed down from adding the Chinese artificial intelligence startup DeepSeek, the Chinese memory chip manufacturer CXMT, and more than 100 other companies to its trade blacklist of entities deemed a threat to U.S. national security, according to two sources familiar with the matter.

This move by the Trump administration aims to avoid escalating tensions with Beijing, as the sources indicate that the vast majority of the companies on the list are Chinese.

As reported by Karen Freifeld of Reuters, the U.S. has held off adding China’s AI startup DeepSeek, memory chipmaker CXMT and more than 100 other companies flagged as national security risks to a trade blacklist, according to two people familiar with the matter, as the Trump administration tries to avoid escalating tensions with Beijing.

The reported decision marks a significant shift in the administration’s approach to China technology policy, prioritizing fragile trade diplomacy over a sweeping tech sector crackdown.

According to Philip Luck, who analyzes supply chains at the Washington Center for International Studies, the U.S. has not added any new entries to its Entity List since October, marking the longest interval without updates in over ten years.

This freeze marks the longest gap in updates to the US blacklist in over a decade, signaling that the administration is temporarily prioritizing fragile trade diplomacy over a sweeping tech sector crackdown.

Why Did a Joint Interagency Committee Approve DeepSeek and CXMT for the Entity List Last Year?

A joint interagency committee had approved the addition of DeepSeek, CXMT, and other companies to the list last year. Internal documents revealed that an inter-agency committee had already greenlit DeepSeek alongside China’s top memory chip maker Changxin Memory Technologies for immediate inclusion on the Commerce Department’s restrictive entity list.

The proposed wave of sanctions targeted more than 100 entities deemed to be severe risks to American national security, including at least 75 Chinese firms specializing in advanced semiconductors and AI development.

ChangXin Memory Technologies, China’s top memory chipmaker, was designated as a Chinese military company by the Defense Department under the Biden administration, according to Economic Times reporting.

This prior designation likely contributed to the interagency committee’s approval for adding CXMT to the Entity List.

What Allegations Did U.S. Officials Make Against DeepSeek Regarding Military Support and Chip Access?

As reported by Michael Martina of Reuters on June 23, 2025, a senior American official informed Reuters that the AI company DeepSeek is supporting the military and intelligence operations of China.

This official indicated that the Chinese tech startup is attempting to utilize shell companies based in Southeast Asia to obtain high-end semiconductors that are restricted from being shipped to China under U.S. regulations.

During an interview, a senior official from the State Department remarked that

“DeepSeek has willingly and will to provide to China military and operations,”

according to Reuters reporting. This statement was made under the condition of anonymity, allowing the official to discuss sensitive U.S. government information.

The official noted that DeepSeek appears over 150 times in procurement documents related to China’s People’s Liberation Army and other organizations connected to the Chinese defense industry, stating that DeepSeek has delivered technological services to PLA research institutions.

As reported by Silicon UK, the US official also alleged that besides using Southeast Asia shell companies to get around the US export restrictions, DeepSeek also allegedly was trying to access data centres in Southeast Asia to remotely access US chips.

According to the Reuters report, the US official alleged DeepSeek is using workarounds to get access to advanced Nvidia chips, evading export controls, and it has access to “large volumes” of Nvidia’s H100 chip, which has been covered by Washington export restrictions since 2022.

DeepSeek reportedly has access to significant quantities of high-end H100 chips from Nvidia.

How Did Anthropic Detect DeepSeek’s Hijacking Campaign Against Its Cloud AI Platform?

As reported by TechCrunch, Anthropic is accusing three Chinese AI companies of setting up more than 24,000 fake accounts with its Claude AI model to improve their own models. The labs — DeepSeek, Moonshot AI, and MiniMax — allegedly generated more than 16 million exchanges with Claude through those accounts using a technique called “distillation”.

In a blog post published on February 22, 2026, Anthropic stated:

“We have detected large-scale campaigns by three AI labs—DeepSeek, Moonshot, and MiniMax—to unlawfully extract Claude’s functionalities to enhance their own models”.

The blog post further stated: “These labs produced over 16 million interactions with Claude through roughly 24,000 fraudulent accounts, violating our service terms and regional access limitations”.

As reported by CyberScoop, the three distillation campaigns followed a similar playbook, using fraudulent accounts and proxy services to access Claude at scale while evading detection. Anthropic said:

“The volume, structure, and focus of the prompts were distinct from normal usage patterns, reflecting deliberate capability extraction rather than legitimate use”.

DeepSeek was responsible for 150,000 of the exchanges, compared to 3.4 million from Moonshot and 13 million from MiniMax, according to the startup.

Anthropic tracked more than 150,000 exchanges from DeepSeek that seemed aimed at improving foundational logic and alignment, specifically around censorship-safe alternatives to policy-sensitive queries.

The labs targeted Claude’s most differentiated capabilities: agentic reasoning, tool use, and coding, according to Anthropic.

One proxy network alone managed more than 20,000 simultaneous fraudulent accounts, mixing distillation traffic with legitimate requests to evade detection. DeepSeek’s campaign, while smaller at 150,000 exchanges, employed particularly sophisticated techniques—including prompts designed to make Claude articulate its internal reasoning step-by-step, essentially generating chain-of-thought training data on demand.

OpenAI also warned lawmakers that DeepSeek was targeting its models, according to Economic Times reporting.

This year, Anthropic said it identified a campaign by DeepSeek and two other Chinese AI labs to illicitly extract capabilities from its Claude AI platform to improve their own models.

What Is China’s Response to the U.S. Blacklist Plans and How Do Experts View the Decision?

China has consistently opposed the US overstretching the concept of national security and abusing export control measures such as the Entity List to suppress and contain Chinese companies, according to Lin Jian, a spokesperson of China’s Foreign Ministry, who said at a regular press conference on Wednesday.

The remarks were made responding to a media inquiry for comment regarding a foreign media report that the US Commerce Department has held off on adding numerous Chinese companies including DeepSeek and a major chip maker to its restrictive Entity List in order to avoid escalating tensions with China.

According to Global Times, the US reportedly pauses plan to blacklist DeepSeek, CXMT and more than 100 other firms is viewed as

“a positive signal to help further stabilize ties”

by experts. The move comes as the Trump administration seeks to prevent further deterioration in ties with Beijing, according to Strategic News Global.

Despite warnings from security officials, Washington is holding its fire for now. The Trump administration has quietly delayed plans to add the breakout Chinese artificial intelligence startup DeepSeek to an official US trade blacklist, prioritizing fragile trade diplomacy amid rising AI and security tensions.

Background of the DeepSeek Blacklist Development

The development surrounding the potential blacklisting of DeepSeek and CXMT began last year when a joint interagency committee approved their addition to the Commerce Department’s Entity List.

The Entity List is a trade blacklist maintained by the U.S. Commerce Department that restricts American companies from exporting certain technologies to listed entities without special government approval.

In June 2025, approximately one year before the recent decision to back down, a senior U.S. State Department official publicly told Reuters that DeepSeek had supported Chinese military and intelligence operations.

This official alleged that the startup had attempted to use shell companies in Southeast Asia to illegally access advanced U.S. chips that cannot be shipped to China under U.S. export rules.

The allegations against DeepSeek gained significant attention when Anthropic publicly announced in February 2026 that it had detected a hijacking campaign by DeepSeek and two other Chinese AI labs.

The company revealed that the labs had conducted industrial-scale distillation attacks using 24,000 fake accounts to steal Claude AI capabilities through over 16 million fraudulent API exchanges.

CXMT, also known as ChangXin Memory Technologies, had earlier been designated as a Chinese military company by the Defense Department under the Biden administration before the Trump administration’s interagency committee approved its addition to the Entity List.

The Entity List has seen minimal updates recently, with no new entries added since October 2025, marking the longest interval without updates in over ten years.

This pause in blacklist additions sets the context for the Trump administration’s decision to hold off on adding the 100+ Chinese firms, including DeepSeek and CXMT.

Prediction: How This Development Can Affect U.S. AI Companies, Technology Sector, and National Security Stakeholders

This development can affect multiple audiences in significant ways. For U.S. AI companies like Anthropic and OpenAI, the decision to delay blacklisting DeepSeek means their Chinese competitors will continue operating without export restrictions on advanced U.S. chips.

This could allow DeepSeek to maintain access to “large volumes” of Nvidia’s H100 chips, which have been covered by Washington export restrictions since 2022.

The technology sector may experience continued uncertainty regarding U.S.-China technology relations.

As reported by Strategic News Global, the move comes as the Trump administration seeks to prevent further deterioration in ties with Beijing, suggesting that trade diplomacy remains fragile. This temporary prioritization of diplomacy over tech crackdown could mean U.S. technology companies face ongoing ambiguity about future export control policies.

For national security stakeholders, the decision raises concerns about whether DeepSeek will continue its alleged support of Chinese military and intelligence operations. The State Department official previously noted that DeepSeek appears over 150 times in procurement documents related to China’s People’s Liberation Army.

Without blacklisting, the company may continue accessing U.S. technology through the workaround methods it allegedly employed, including Southeast Asian shell companies and remote access to data centers.

U.S. AI companies that have detected targeting by DeepSeek, including Anthropic and OpenAI, may need to strengthen their security measures against distillation attacks and fraudulent account campaigns.

The 16 million fraudulent exchanges and 24,000 fake accounts demonstrated by DeepSeek, Moonshot, and MiniMax indicate sophisticated capabilities that could continue threatening U.S. AI platforms.

For investors and analysts monitoring U.S.-China technology tensions, the longest gap in Entity List updates in over a decade signals a temporary shift in administration priorities. However, as noted in YouTube reporting on the developments,

“this is just temporary. We don’t know how they’re going to move forward,”

suggesting the blacklisting could resume if diplomatic tensions escalate.

Chinese companies on the proposed list, including at least 75 firms specializing in advanced semiconductors and AI development, will continue operating without the restrictions that would have limited their access to U.S. technology.

This could accelerate China’s AI and semiconductor development in the short term, potentially affecting long-term competitive dynamics in these critical technology sectors.