Ethical Hacking News
U.S. cybersecurity and intelligence agencies have accused China-based AI firms of conducting "systematic extraction" of proprietary functionalities and capabilities from American frontier models through distillation attacks, posing a significant threat to AI security and potentially compromising sensitive data.
China-based AI firms have been accused of engaging in "systematic extraction" of proprietary functionalities from American AI models through distillation attacks.These firms have been extracting billions of tokens from U.S. frontier AI models, including variants of Anthropic Claude, OpenAI GPT, Google Gemini, and SpaceXAI Grok.The extraction is done through various methods, including APIs, remote cloud providers, and third-party aggregators that obfuscate user metadata.The China-based AI companies are achieving cost savings by distributing the operations across multiple providers and platforms to avoid detection.The U.S. government has warned of the severity of the situation and recommended that U.S. AI companies implement detection and mitigation measures.The incident highlights the need for a coordinated response against sophisticated, well-resourced adversaries and the importance of AI security.
The recent allegations made by U.S. cybersecurity and intelligence agencies against China-based artificial intelligence (AI) firms have shed light on a disturbing trend in the AI landscape. The accusations, which have been corroborated by experts, reveal that these firms have been engaged in "systematic extraction" of proprietary functionalities and capabilities from American frontier models through distillation attacks.
This phenomenon, which has been described as occurring at an industrial scale, forms the "core" of China's AI development strategy. The targeted distillation activities, which have been ongoing since at least late 2024, have resulted in the extraction of billions of tokens across millions of exchanges/requests from U.S. frontier AI models. The models in question include variants of Anthropic Claude, OpenAI GPT, Google Gemini, and SpaceXAI Grok.
The distillation requests are routed through various methods to gain unauthorized access to the U.S. models, violating their terms of use. These include application programming interfaces (APIs), remote cloud providers, and third-party aggregators that obfuscate user metadata to avoid detection. It is worth noting that U.S. frontier models are officially restricted and not offered in China, forcing Chinese developers to rely on domestic systems or alternative access methods like virtual private networks, obfuscated accounts, and automated agents to bypass these geographic controls.
Furthermore, the China-based AI companies deliberately take steps to distribute these operations across multiple providers and platforms to fly under the radar, focusing on distilling the best capabilities and proprietary features of each U.S. frontier model to train their own models. This sophisticated approach has enabled the firms to achieve cost savings for their industrial-scale distillation campaigns through bulk procurement of the U.S. AI companies' premium subscriptions shared across teams of developers.
The joint advisory released by the National Security Agency (NSA), the Cybersecurity and Infrastructure Security Agency (CISA), and the Federal Bureau of Investigation (FBI) has highlighted the severity of the situation. The agencies have warned that China-based AI companies that conduct industrial-scale distillation against U.S. AI models see significantly shorter AI development timelines and reduced financial expenditures in training a frontier model.
However, the experts argue that this threat should not be dismissed as irrelevant to businesses not directly associated with frontier AI models. The exposure of model access to customers or partners makes API keys and service accounts valuable targets, with abuse of access to those models appearing as legitimate. The agencies have recommended that U.S. AI companies implement comprehensive detection and mitigation measures, subtly alter responses for suspected malicious distillation attempts, and correlate activity across model providers, cloud platforms, and API aggregators to reveal distributed campaigns.
The incident highlights the need for a coordinated response against sophisticated, well-resourced adversaries. As Ismael Valenzuela, vice president of Labs, Threat Research and Intelligence at Arctic Wolf, noted, "Attackers deploy proxy infrastructure to orchestrate large-scale automated attacks, rotating queries across thousands of compromised credentials and fraudulent accounts across different product channels to obscure their origin and bypass standard security controls."
The security bulletin has also been interpreted as an abuse of legitimate access, while stressing the need for a coordinated response against sophisticated, well-resourced adversaries. The incident serves as a stark reminder of the importance of AI security and the need for robust detection and mitigation measures to protect against these types of threats.
Related Information:
https://www.ethicalhackingnews.com/articles/Exposing-the-Shadowy-World-of-Distilled-AI-The-Growing-Concern-of-Chinas-AI-Firms-ehn.shtml
https://thehackernews.com/2026/09/us-agencies-accuse-china-ai-firms-of.html
Published: Wed Sep 9 07:00:48 2026 by llama3.2 3B Q4_K_M
