US cybersecurity agencies including the NSA, FBI, and CISA have accused six Chinese AI firms of conducting industrial-scale campaigns to extract capabilities from American models. The targeted companies include DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI. Officials allege these firms used a technique known as distillation to access models from Anthropic, OpenAI, Google, and xAI starting in late 2024.
The agencies claim these operations involved buying premium accounts and routing traffic through transfer stations to avoid detection, aiming to capture reasoning and coding skills quickly. China has rejected these accusations as false smears, arguing that distillation is a neutral technique and that its AI progress stems from self-reliance. The US administration plans to discuss these safety issues during upcoming talks between President Trump and President Xi Jinping.
In response to the warnings, US officials are urging American AI companies to treat model outputs as a security risk. Agencies recommend secretly degrading answers sent to suspected malicious traffic by varying responses and limiting reasoning depth to make stolen data less useful. This diplomatic and security tension has already impacted markets, with Alibaba stock falling 2.6% after the US labeled its AI methods a threat.
Separately in the robotics sector, Unusual Machines announced an additional $20 million investment in XTEND AI Robotics, bringing total funding to $27.5 million. This capital supports the commercialization of autonomous drone technology capable of performing complex tasks without human assistance.
Key Takeaways
- US agencies accuse six Chinese firms—DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI—of maliciously copying American AI models via distillation.
- The alleged data extraction campaigns began in late 2024 and targeted models from Anthropic, OpenAI, Google, and xAI.
- China rejects the US claims, stating its AI development relies on self-reliance and calling the accusations false smears.
- US officials warn that these operations likely occurred with the knowledge of the Chinese government to save costs and speed up development.
- Agencies recommend American AI providers degrade responses sent to suspected malicious traffic to limit the utility of stolen data.
- Alibaba stock dropped 2.6% following US accusations that its AI development methods pose a security threat.
- Alibaba shares have fallen 10% in the past month amid increased US scrutiny of Chinese technology companies.
- US and Chinese leaders plan to discuss AI safety issues during an upcoming meeting in late September.
- Unusual Machines invested an additional $20 million in XTEND AI Robotics, raising total funding to $27.5 million.
- The new funding for XTEND AI will support the commercialization of autonomous drone technology for complex tasks.
China rejects US claims of malicious AI copying
The US government accused Chinese AI companies of stealing technology from American models through a process called distillation. Agencies like the FBI and NSA stated that firms such as DeepSeek and Alibaba extracted billions of tokens since late 2024. China responded by calling these accusations false and urged the US to stop making smears against its technology. The two nations plan to discuss these AI safety issues during upcoming talks between President Trump and President Xi Jinping.
NSA warns Chinese firms steal US AI model data
The National Security Agency warned that Chinese companies are extracting capabilities from US AI models on a large scale. The report names DeepSeek, Moonshot AI, and Alibaba as firms that used millions of exchanges to get data from models like Claude and GPT. Officials believe these operations happened with the knowledge of the Chinese government to save money and speed up development. The agencies recommend that US companies improve their detection systems to stop these unauthorized access attempts.
US accuses six Chinese AI firms of malicious copying
US officials accused six Chinese AI companies of maliciously copying technology from American firms using a technique known as distillation. The targeted companies include DeepSeek, Moonshot AI, and Alibaba, which accessed models from Anthropic, OpenAI, Google, and SpaceX. The US administration made these claims just before President Trump prepares to meet with Chinese President Xi Jinping in late September. China rejected the claims, stating its AI progress comes from self-reliance and calling the US accusations false.
US agencies warn Chinese AI firms extract advanced models
US agencies accused six Chinese AI firms of extracting billions of tokens from US models to copy advanced capabilities quickly. The advisory details how companies like DeepSeek ran organized campaigns against models such as Claude and GPT starting in late 2024. These firms used methods like buying premium accounts and routing traffic through transfer stations to avoid detection. The goal was to capture reasoning skills and coding abilities that usually take years to develop.
US agencies accuse Chinese AI firms of industrial extraction
US cybersecurity agencies accused six Chinese AI companies of conducting industrial-scale campaigns to extract capabilities from American frontier models. The named firms include DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI, which targeted models from Anthropic, OpenAI, Google, and xAI. The agencies allege these operations began in late 2024 and were likely done with the awareness of the Chinese government. China rejected the claims, arguing that distillation is a neutral technique and accusing the US of applying double standards.
CISA warns Chinese firms steal US AI skills
The US Cybersecurity and Infrastructure Security Agency warns that Chinese AI companies are stealing capabilities from American models. These firms use a technique called knowledge distillation to copy reasoning and coding skills at an industrial scale. Agencies believe this activity happens with the knowledge of the Chinese government and targets models from companies like OpenAI and Google. US officials urge American AI companies to take immediate steps to protect their platforms from these campaigns.
US accuses six Chinese firms of malicious AI copying
US law enforcement officials accused six Chinese AI companies of maliciously copying American technology through a process known as distillation. The targeted companies include DeepSeek, Moonshot AI, and Alibaba, which used US models to train their own products faster. Officials state this activity likely occurred with the awareness of the Chinese government and involved models from Anthropic and SpaceX. This accusation could complicate diplomatic relations ahead of a planned meeting between the leaders of the US and China.
US agencies tell AI firms to hide answers from spies
The NSA, CISA, and FBI issued a joint advisory asking American AI developers to treat model outputs as a security risk. Agencies warn that Chinese companies are running industrial-scale campaigns to steal data from US models. The officials recommend that providers secretly degrade answers sent to suspected malicious traffic to stop the theft. This strategy involves varying responses and limiting reasoning depth to make the stolen data less useful.
Alibaba stock drops after US calls AI methods a threat
Alibaba stock fell 2.6% after the US government labeled its AI development methods as a security threat. The US has become increasingly wary of Chinese technology companies and their influence in the American market. This move is part of a series of actions aimed at limiting the reach of Chinese tech firms. Analysts note that Alibaba shares have been volatile and have already fallen 10% in the past month.
Unusual Machines adds $20 million to XTEND AI Robotics
Unusual Machines announced an additional $20 million investment in XTEND AI Robotics, bringing the total funding to $27.5 million. This strategic investment supports the development of XTEND AI's autonomous drone technology. The new funds will help commercialize solutions that allow drones to perform complex tasks without human help. Unusual Machines CEO stated that the partnership aims to revolutionize the drone industry through innovation.
Sources
- China hits back at US claims of ‘malicious’ AI distillation ahead of planned Trump-Xi talks
- The National Security Agency warns that Chinese companies are extracting capabilities from US AI models
- US accuses six Chinese AI firms of 'malicious' copying
- US Agencies Warn Chinese AI Firms Are Extracting Advanced AI Models
- U.S. Agencies Accuse Chinese AI Companies of Industrial-Scale Extraction From American Models
- Chinese AI firms are siphoning capabilities from American models, CISA warns
- US accuses Chinese AI firms of ‘malicious’ copying of AI technology
- US security agencies tell AI firms to degrade answers to suspected Chinese distillation traffic in secret
- Alibaba Drops 2.6% as U.S. Turns AI Distillation Into a Security Threat
- Unusual Machines Invests an Additional $20 Million in XTEND AI Robotics, Bringing Total Investment to $27.5 Million
Comments
Please log in to post a comment.