Artificial Intelligence – US Firm Raises Alarm Over Alleged AI Model Misuse
Artificial Intelligence – A leading American artificial intelligence company has raised serious allegations against three prominent Chinese startups, claiming that its proprietary technology may have been improperly used to enhance competing AI systems.
The claims, highlighted in a recent CNN Business report, center around accusations by Anthropic, a US-based AI firm, against Chinese companies DeepSeek, MiniMax, and Moonshot AI. According to the report, Anthropic believes these firms may have extracted capabilities from its Claude AI model through unauthorized methods.
Allegations of Unauthorized Model Training
Anthropic has suggested that the suspected misuse involved a technique known as “distillation,” where outputs from an advanced AI model are used to train another system. While distillation itself is not inherently illegal, the company argues that in this case it may have been carried out without proper authorization.
The report indicates that the alleged activity involved the creation of approximately 24,000 fake user accounts. These accounts were reportedly used to generate more than 16 million interactions with the Claude model, which could then be analyzed and replicated to improve other AI systems.
Concerns Over Safety and Misuse
Anthropic expressed concern that AI models developed using such methods might not include the same safety protocols built into its own systems. The company warned that this could potentially open the door to misuse in areas such as cyberattacks or even the development of harmful biological tools.
It also highlighted broader geopolitical risks, suggesting that advanced AI tools lacking proper safeguards could be leveraged for mass surveillance, disinformation campaigns, or offensive cyber operations. According to the company, the situation presents a limited window for policymakers to respond effectively.
Industry Response and Lack of Comments
The report noted that CNN has reached out to DeepSeek, MiniMax, and Moonshot AI for their responses, but no official statements had been made at the time of publication.
These companies have recently gained attention in China’s rapidly expanding AI sector, with some analysts referring to them as rising leaders in the field. Their models are currently ranked among the top performers on widely followed AI benchmarking platforms.
Debate Over Export Controls
The issue has also reignited discussions around the effectiveness of US export controls on advanced semiconductor technology. Some observers argue that the rapid progress of Chinese AI firms suggests these restrictions may not be fully effective.
However, Anthropic maintains a different view. The company stated that such incidents actually demonstrate the importance of export controls, emphasizing that sustained development of cutting-edge AI systems still depends heavily on access to high-performance computing hardware.
Previous Similar Allegations
This is not the first time concerns of this nature have been raised within the industry. Earlier, OpenAI had also accused DeepSeek of benefiting from technologies developed by leading US AI labs without adequate attribution or licensing.
These recurring claims reflect growing tensions in the global AI race, particularly between the United States and China, where technological leadership is increasingly seen as a matter of national interest.
Government Classification and Clarifications
In a related development, Anthropic has recently been designated as a “Supply Chain Risk” entity by the US government. The classification applies specifically to the use of its AI models in certain defense-related contracts.
The company clarified that this designation does not restrict general commercial use of its technology. It applies only in specific contexts involving Department of Defense agreements.
Anthropic’s leadership has also been in the spotlight recently, with its CEO issuing a public apology for earlier remarks concerning former US President Donald Trump.