Why is Anthropic calling for U.S. Congressional action?

Anthropic is appealing to Congress because it views the alleged cloning as a significant act of intellectual property theft and a national security threat. The company believes that such large-scale extraction undermines U.S. technological leadership in AI and that existing legal frameworks may not be sufficient to address this new form of digital appropriation. Congressional action could lead to new regulations, investigations, or economic penalties against Alibaba.

What are the national security implications of AI model cloning?

The national security implications stem from the strategic importance of advanced AI. If a foreign power can quickly replicate cutting-edge AI developed by U.S. firms, it could gain military, intelligence, or economic advantages without the corresponding research investment. The U.S. government has already indicated its concern by imposing export controls on Anthropic's latest models, suggesting that the capabilities of these AIs are deemed critical to national interests.

Has Alibaba responded to Anthropic's accusations?

As of June 26, 2026, Alibaba has not yet issued a public comment or response regarding Anthropic's allegations. Media outlets, including the BBC, have reportedly contacted Alibaba for comment.

Image: courtesy of Ars Technica

techJune 26, 2026By Veridact EditorialUpdated Jun 26

Anthropic Demands Action Against Alibaba Over Alleged Massive AI Cloning Attack

Anthropic, a leading U.S. artificial intelligence firm, has publicly accused Chinese tech giant Alibaba of orchestrating the largest-known cloning attack on its advanced Claude AI model. The company claims Alibaba used 25,000 fraudulent accounts to conduct over 28.8 million exchanges, illicitly extracting Claude's capabilities. Anthropic has urged the U.S. Congress to penalize Alibaba and other companies engaged in similar activities, citing national security concerns and the need to protect American technological leadership. Alibaba has not yet issued a public response to these serious allegations.

Outlook

The accusation by Anthropic marks a significant escalation in the ongoing tech and intellectual property disputes between the United States and China, particularly concerning advanced AI. The immediate focus will be on Alibaba's response, or lack thereof, and whether U.S. lawmakers will heed Anthropic's call for action. Given the scale of the alleged attack and the national security context, a quiet resolution appears unlikely. The U.S. government, already sensitive to technology transfers, is now under pressure to demonstrate a firm stance against what Anthropic describes as illicit extraction of its AI capabilities. This could manifest in heightened regulatory scrutiny, investigations, or even sanctions targeting Alibaba or its affiliates. For other AI developers, this incident serves as a stark warning about the vulnerabilities of advanced models to persistent, large-scale extraction efforts.

We can expect intense behind-the-scenes lobbying from both sides, with Anthropic pushing for aggressive measures and Alibaba likely preparing a robust defense or a denial of the allegations. The incident could also prompt other U.S. AI firms to review their own security protocols and engagement policies with foreign entities.

Background

The accusations against Alibaba come at a particularly sensitive time for the global AI industry and U.S.-China relations. Anthropic's 'Claude' models are considered among the most advanced large language models (LLMs) in the world, capable of complex reasoning, content generation, and sophisticated interaction. The company recently released its latest models, Fable 5 and Mythos 5, which the U.S. government subsequently restricted from access by foreign nationals through an export control directive. This directive, issued by the Trump administration earlier this month, cited 'national security authorities' without specifying further details, indicating a clear concern over the potential misuse or strategic appropriation of cutting-edge AI.

Anthropic's letter, dated June 10 and addressed to Senators Tim Scott and Elizabeth Warren, explicitly links the alleged cloning attack to these broader national security concerns. The company argues that such large-scale extraction efforts undermine the significant investment and innovation made by U.S. companies in AI, effectively allowing foreign rivals to gain competitive advantages without the corresponding research and development costs. The specific mention of 25,000 accounts and over 28.8 million exchanges provides a concrete measure of the alleged effort, painting a picture of a systematic and persistent campaign rather than isolated incidents. The method of 'cloning' an AI model typically involves sending a massive volume of queries to the target model and then using the responses to train a new model that mimics its behavior and capabilities. This process, often referred to as 'model distillation' or 'knowledge extraction,' can allow a rival to create a functionally similar AI without having access to the original source code or proprietary training data, effectively circumventing intellectual property protections.

Precedents

The alleged cloning attack by Alibaba fits into a long-standing pattern of intellectual property disputes and technology transfer concerns between the United States and China. For decades, U.S. companies have accused Chinese entities of various forms of IP theft, ranging from industrial espionage to reverse engineering and cyberattacks. High-profile cases have involved everything from manufacturing secrets to software algorithms, frequently leading to trade disputes, sanctions, and legal battles.

In the realm of technology, especially advanced sectors like semiconductors, telecommunications, and now AI, the U.S. government has increasingly framed these issues as national security threats. The export control directives placed on Huawei and other Chinese tech firms in recent years, as well as the recent restrictions on Anthropic's own models, illustrate a clear policy shift towards safeguarding critical technologies. The concern is that if foreign adversaries can 'clone' advanced AI models, they can rapidly close technological gaps and potentially use these capabilities for military, surveillance, or economic advantage.

However, proving and prosecuting digital intellectual property theft, particularly across international borders and involving complex AI models, remains a challenge. The 'cloning' process, while ethically questionable from a competitive standpoint, often operates in a legal grey area, as it involves interacting with a publicly accessible API rather than directly stealing code or data. Past attempts to penalize foreign companies for such activities have often relied on trade laws, sanctions, or diplomatic pressure, with mixed results. The precedent suggests that while the U.S. government is likely to take these accusations seriously, the path to 'punishment' for Alibaba may be complex and protracted.

This incident is far more than a simple corporate disagreement; it represents a flashpoint in the global race for AI supremacy and the broader geopolitical competition between the U.S. and China. For Anthropic, a company that has invested billions in developing its cutting-edge AI, the alleged cloning directly threatens its competitive advantage and the very value of its intellectual property. If sophisticated models can be effectively replicated through large-scale querying, it erodes the incentive for costly frontier AI research and development.

On a national level, the accusations touch on core U.S. strategic interests. The ability of a foreign power to rapidly acquire or replicate advanced AI capabilities without independent innovation is seen as a direct challenge to U.S. technological leadership and national security. The U.S. government's recent export controls on Anthropic's advanced models underscore this concern, indicating that AI is now viewed as a critical technology requiring stringent protection. Should Congress act on Anthropic's appeal, it could set a powerful new precedent for how the U.S. responds to digital IP theft in the AI era, potentially leading to more aggressive enforcement mechanisms and further decoupling of tech ecosystems.

For the broader AI industry, this event highlights the urgent need for clearer international norms and legal frameworks around AI model protection. The current legal landscape is ill-equipped to handle the nuances of 'cloning' via API interaction, which differs from traditional software piracy. The outcome of this dispute could shape future investment in AI, influence the openness of AI research, and redefine the boundaries of competitive practices in the most critical technology of the 21st century. It could also force companies to reconsider how they deploy and protect their most valuable AI assets.

Scenarios

Analysis

One immediate outcome could be a significant diplomatic and economic response from the U.S. government. Given the explicit call for action from Anthropic to Congress and the existing national security directive on AI, U.S. lawmakers may initiate investigations, hold hearings, or even consider targeted sanctions against Alibaba. This could involve restricting Alibaba's access to U.S. technologies or financial markets, similar to measures taken against other Chinese tech firms. Such actions would not only aim to penalize Alibaba but also to send a strong message to other foreign entities contemplating similar activities.

Conversely, Alibaba could issue a forceful denial, claiming the accusations are unfounded, a misunderstanding of their data usage, or a mischaracterization of their AI development practices. They might argue that their activities fall within accepted norms of competitive intelligence gathering or that the 'cloning' is a natural outcome of independent research. This could lead to a protracted legal or public relations battle, with both companies presenting their evidence and interpretations. This scenario would likely result in increased scrutiny of AI data practices globally, but without immediate, decisive punitive action from the U.S. government, it might leave the underlying issue of AI IP protection largely unresolved in the short term, pushing the burden onto individual companies to harden their models against extraction.

Timeline

2026-06-10

Anthropic Sends Letter to Congress

Anthropic sends a letter to U.S. Senators Tim Scott and Elizabeth Warren, accusing Alibaba of the largest-known cloning attack on its Claude AI model and urging congressional action.

2026-06-XX

U.S. Export Control Directive Issued

The Trump administration issues an export control directive, ordering Anthropic to suspend access to its latest Claude models (Fable 5 and Mythos 5) by any foreign national, citing national security authorities. (Specific date not provided, but occurred earlier in June).

2026-06-25

Accusations Go Public

News reports emerge detailing Anthropic's accusations against Alibaba, highlighting the scale of the alleged attack (25,000 accounts, over 28.8 million exchanges).

2026-06-26

Alibaba Silence Continues

As of this date, Alibaba has not yet issued a public comment or response to Anthropic's allegations.

Frequently Asked Questions

In this context, 'cloning' an AI model refers to a process where an entity sends a massive number of queries or prompts to an existing AI model, like Anthropic's Claude, and then uses the responses to train its own new AI model. This new model is designed to mimic the behavior, capabilities, and knowledge of the original, effectively creating a functional copy without needing access to the original's internal architecture, source code, or proprietary training data.

Discussion

Be the first to share your thoughts.