Anthropic Accuses Alibaba of Major AI Extraction Attempt

Recently, Anthropic has made serious allegations against the Chinese tech corporation Alibaba, claiming it conducted the most significant known attempt to hijack its AI technology through a method called a “distillation” attack.

On June 10, Anthropic sent a letter to the U.S. Senate Committee on Banking, Housing, and Urban Affairs, pointing fingers at Alibaba for what it described as an illegal extraction of its AI capabilities. This letter, which was shared with Republican Senators Tim Scott and Elizabeth Warren, was confirmed by CNBC this week.

In the letter, Anthropic claimed that Alibaba executed the largest distillation attack against the firm. Distillation is a technique where the performance data generated by a robust AI model is used to develop a smaller, less powerful version. Between April 22 and June 5, Anthropic reported that individuals linked to Alibaba managed to interact with its AI models a staggering 28.8 million times through around 25,000 fake accounts.

An Anthropic representative noted, “We think it’s crucial to combat illegal distillation through teamwork between the government and the industry. We are committed to partnering with Congress and the administration to uphold the U.S.’s leadership in AI.” Meanwhile, Alibaba has yet to respond to inquiries from CNBC.

This isn’t the first instance of such claims from Anthropic. Back in February, the company reported discovering three large-scale distillation efforts coming from other Chinese AI organizations: DeepSeek, Moonshot, and MiniMax. At that time, Anthropic remarked that the intensity and sophistication of these campaigns were escalating, urging collaboration across the AI sector, cloud service providers, and regulatory bodies.

In related news, Alibaba is currently embroiled in a lawsuit against the Department of Defense, aiming to have its designation as a “Chinese military company” removed.

The Department of the Army maintains a list of firms providing services to the People’s Liberation Army, known as the “1260H List.” Since 2021, this list has evolved to reflect the merging of China’s civilian businesses with its military objectives. On June 8, several major companies, including Alibaba, were added to this list. Opponents of the designation, including the Chinese government, claim it is arbitrary and unjust.

The Chinese Embassy in Washington responded to the recent updates on the list, stating, “The United States should eliminate unfair practices and foster a just, equitable, and non-discriminatory environment for Chinese firms.”

As the U.S. contemplates how to leverage AI effectively, conservatives face challenges on two fronts. Domestically, left-leaning factions in Silicon Valley are vying for control over AI, while internationally, China is attempting to assert its influence as well. Wynton Hall, social media director at Breitbart News, emphasizes these challenges in his bestseller, Code Red: Left, Right, China, and the Race to Control AI, which discusses how the MAGA movement can navigate the AI landscape without conceding power to either tech elites or foreign adversaries.