US-based artificial intelligence startup Anthropic has accused Chinese technology giant Alibaba of using thousands of fake accounts to gain unauthorized access to its proprietary AI model, Claude.

According to reports by Bloomberg, the Financial Times, and Reuters, which cited an official letter sent by the company as well as informed sources, the allegations were formally communicated to US senators and White House officials.

In the letter, Anthropic asserted that activities conducted by operators linked to Alibaba targeted the most valuable capabilities of the Claude model, including its software development functionalities.

The company characterized the incident as the largest attempt to date by a Chinese firm to leverage pioneer US artificial intelligence technologies for its own benefit.

Twenty-nine million suspicious transactions in three months

According to data compiled by Anthropic, approximately 29 million transactions linked to the Claude model were executed through roughly 25,000 fake accounts between April and June.

The company noted that Alibaba and other China-based firms systematically exploit leading US technologies to develop their own chatbots.

In the letter, as reported by Bloomberg, Anthropic officials evaluated the process, stating:

“These attacks, carried out through distillation methods, were executed systematically and on an industrial scale to illegally copy advanced US AI technologies from leading laboratories, bypassing training and research-and-development costs to present them as their own products.”

The Financial Times pointed out that the distillation method is widely used in the technology sector to train cheaper and smaller versions of artificial intelligence models.

However, US officials are concerned that the use of this method by Chinese competitors to develop their own models could carry serious national security implications.

Call to Congress to close loopholes

According to the Financial Times report, Anthropic urged the US Congress to close legal loopholes that allow Chinese AI firms to access advanced US technologies, and to penalize the Chinese companies responsible for these cyber activities.

The company also stated that Alibaba pursued this activity brazenly, even after the White House issued a directive emphasizing the need to prevent intellectual property theft at artificial intelligence firms.

As reported by Reuters, Anthropic emphasized in its letter that it supports the Washington administration’s efforts to combat cyberattacks.

On June 13, Anthropic announced that the US government had mandated blocking access to its most advanced AI models, Fable 5 and Mythos 5, for all foreign users who are not US citizens.

Subsequently, David Sacks, a US investor and co-chair of the President’s Council of Advisors on Science and Technology, explained that the decision was taken following the detection of possibilities that the built-in security mechanisms of the models could be bypassed.

Uncertainty in the global arms market, driven by the United States reassessing its relationships with allies and a broad rearmament drive across many countries, is creating major commercial opportunities for South Korea. According to an analysis published by Politico, Seoul has become the world’s fastest-growing supplier of military equipment.

The report said that large-scale conflicts around the world have created urgent demand for weapons as countries seek both to support allies and strengthen their own defenses against potential future confrontations. At the same time, changes in the US role within the global arms market have opened new opportunities for South Korean manufacturers. Statements and policy decisions by US President Donald Trump regarding NATO have led allies to question Washington’s reliability in times of crisis, increasing uncertainty across the global market. In addition, the diversion of a large share of US weapons supplies to the Middle East because of ongoing conflicts has placed further strain on already overstretched supply chains.

European countries increase purchases from South Korea

Faced with what Politico described as the Trump administration’s more distant approach toward allies, European countries in particular have accelerated arms purchases from South Korea. The publication noted that Seoul’s growing influence as a supplier has been driven largely by major defense contracts signed with Poland.

Following the outbreak of the conflict in Ukraine, several Eastern European capitals, including Warsaw, transferred portions of their military inventories to Kyiv, relying on German support to replenish their arsenals. However, Berlin’s slow pace in replacing allied stockpiles generated frustration across the region.

South Korea emerged as an alternative supplier during this period and became a reliable source of military equipment for Eastern European countries. Poland became Seoul’s largest customer through a $13.7 billion agreement covering the purchase of tanks, rocket launchers, self-propelled howitzers and other military equipment.

“We were originally preparing against North Korea, but now we are ready to provide these solutions to customers around the world,” said Choo Hyung-kim, head of the Security Management Institute, a defense analysis organization affiliated with South Korea’s National Assembly.

Lack of political baggage gives Seoul an advantage

Politico reported that one of the greatest advantages enjoyed by South Korean defense companies is the absence of the “political baggage” associated with major arms exporters such as the United States, China, Russia and Israel.

According to the figures cited, the combined projected revenue of South Korea’s largest defense companies, including Hanwha Group, Hyundai Rotem, LIG Nex1 and Korea Aerospace Industries, is expected to reach approximately $37 billion in 2026. That would represent a fourfold increase from their combined revenues in 2021.

Meanwhile, an official from the office of former South Korean President Yoon Suk-yeol told the Yonhap news agency in 2024 that the scale of any weapons shipments to Ukraine would depend on Russia’s approach to its relationship with North Korea. Seoul later clarified that it had no plans to provide ammunition directly to Ukraine.

Chinese artificial intelligence startup DeepSeek has raised more than 50 billion yuan ($7.4 billion) in its first funding round. According to Reuters, citing The Information, the company’s valuation has surpassed $50 billion.

The Wall Street Journal (WSJ) reported that the capital will be used to support the costly development of advanced artificial intelligence technologies.

According to the newspaper, citing sources familiar with the matter, investors valued the company at more than $50 billion. The valuation makes DeepSeek the most valuable AI startup in China.

DeepSeek founder Liang Wenfeng reportedly owned about 90% of the company before the funding round. Liang is said to have contributed roughly $3 billion during the fundraising process, making him the largest participant in the round.

According to Reuters, the transaction was structured in an unusual way that allows Liang to retain control of the company.

Rather than investing directly in DeepSeek, investors were required to invest through a limited partnership managed by a senior executive of the startup. Under the arrangement, investors were not granted voting rights. The report also said restrictions were placed on the use of invested funds for a period of five years.

The sole exception was the China National Artificial Intelligence Industry Investment Fund. The fund reportedly invested approximately $150 million directly in DeepSeek, allowing it to retain both voting rights and full discretion over its stake.

Other major investors in the funding round included Tencent, which invested approximately $1.5 billion, and Contemporary Amperex Technology, which invested about $740 million.

Bloomberg previously described the transaction as one of the largest fundraising rounds undertaken by a Chinese startup. According to the agency, the investment marks a new stage in the efforts of leading Chinese AI companies to compete with their US rivals.

DeepSeek told prospective investors that it would prioritize foundational and transformative AI research over short-term commercialization.

Based in the Chinese city of Hangzhou, DeepSeek emerged as one of Beijing’s most prominent AI companies after unveiling a more powerful and lower-cost model more than a year ago. The WSJ reported that interest surrounding the company has accelerated AI adoption in China and increased investor appetite for domestic startups.

Liang Wenfeng has previously said he intends to continue developing open-source AI models and ultimately aims to achieve artificial general intelligence (AGI). According to Bloomberg, the strategy continues an approach that has contributed to the spread of open models and influenced companies across China’s AI market, including Alibaba’s Qwen platform.

Bloomberg added that while global rivals such as OpenAI and Anthropic are exploring public offerings and revenue-generation strategies, DeepSeek has maintained its “research first” approach.