Anthropic Claims Chinese AI Companies are “Distilling” Claude

Anthropic has accused three Chinese AI firms of misusing its Claude chatbot to strengthen their own systems, saying the activity violated its policies and highlights the need for tighter controls on advanced chip exports. 

In a recent statement, the San Francisco-based company said MiniMax, DeepSeek, and Moonshot generated more than 16 million combined exchanges with Claude through approximately 24,000 fraudulent accounts. 

The company said the firms relied on a method known as distillation, a practice in which a smaller or less advanced model is trained using responses produced by a more capable system. Anthropic warned that such efforts are becoming more organized and technically advanced. The company described the activity as escalating in both scale and complexity, adding that the opportunity to respond is limited and the implications go beyond any one business or country. 

Earlier this month, OpenAI raised similar concerns. In a memo to U.S. legislators, the maker of ChatGPT cautioned that Chinese AI startup DeepSeek was attempting to copy leading American models and use them to accelerate its own development. 

Anthropic’s statement noted that the models trained through unauthorized distillation may not include the same safety protections as the originals. Without built-in safeguards, the company argued, these systems could pose national security concerns. The risks increase if such models are released publicly, since open distribution would allow advanced capabilities to spread beyond regulatory oversight. 

The company recently completed a major funding round, securing $30 billion and reaching a reported valuation of $380 billion. It said the alleged misuse of its technology strengthens the argument for limiting access to high-performance chips used in AI training. Restricting chip exports, Anthropic said, not only slows direct model development but also reduces the scale of improper copying efforts. 

DeepSeek allegedly focused on enhancing reasoning skills across a range of tasks and developing alternatives capable of handling politically sensitive questions in ways that bypass typical content controls. Moonshot, according to Anthropic, concentrated on agent-style reasoning, tool integration, programming, and data analysis. 

MiniMax pursued similar goals, targeting coding capabilities, tool coordination, and multi-step task management. Anthropic said it identified MiniMax’s activity while the campaign was still underway, before the Chinese firm released the model it had been developing. 

The company added that when it introduced a new version of Claude during MiniMax’s campaign, the traffic pattern shifted quickly. Within a day, nearly half of MiniMax’s queries were redirected toward the updated system, suggesting an effort to capture improvements from the latest release. 

The concerns expressed by Anthropic regarding unauthorized copying of AI solutions are likely to be of interest to other players in the tech space, such as GlobalTech Corp. (OTC: GLTK) since their own products could at one point be targeted too. 

About AINewsWire

AINewsWire (“AINW”) is a specialized communications platform with a focus on the latest advancements in artificial intelligence (“AI”), including the technologies, trends and trailblazers driving innovation forward. It is one of 75+ brands within the Dynamic Brand Portfolio @ IBN that delivers: (1) access to a vast network of wire solutions via InvestorWire to efficiently and effectively reach a myriad of target markets, demographics and diverse industries; (2) article and editorial syndication to 5,000+ outlets; (3) enhanced press release enhancement to ensure maximum impact; (4) social media distribution via IBN to millions of social media followers; and (5) a full array of tailored corporate communications solutions. With broad reach and a seasoned team of contributing journalists and writers, AINW is uniquely positioned to best serve private and public companies that want to reach a wide audience of investors, influencers, consumers, journalists, and the general public. By cutting through the overload of information in today’s market, AINW brings its clients unparalleled recognition and brand awareness.

AINW is where breaking news, insightful content and actionable information converge.

To receive SMS alerts from AINewsWire, text “AI” to 888-902-4192 (U.S. Mobile Phones Only)

For more information, please visit www.AINewsWire.com

Please see full terms of use and disclaimers on the AINewsWire website applicable to all content provided by AINW, wherever published or re-published: https://www.AINewsWire.com/Disclaimer

AINewsWire
Los Angeles, CA
www.AINewsWire.com
310.299.1717 Office
Editor@AINewsWire.com

AINewsWire is powered by IBN