State Department orders global campaign against Chinese AI model theft
The U.S. State Department has instructed diplomatic posts worldwide to alert foreign governments about alleged intellectual property theft by Chinese AI companies, including DeepSeek, according to an internal cable dated Friday.
The directive tells embassy and consulate staff to brief counterparts on "concerns over adversaries' extraction and distillation of U.S. A.I. models." A separate message was sent to Beijing raising the same concerns with Chinese officials.
How the theft allegedly works
The companies are accused of using model distillation-a technique where smaller AI models are trained on the outputs of larger, more expensive ones. This approach allows competitors to create capable systems at a fraction of the original development cost.
The cable states that distilled models "appear to perform comparably on select benchmarks at a fraction of the cost but do not replicate the full performance of the original system." It also claims the companies deliberately remove security safeguards and neutrality mechanisms from the resulting models.
DeepSeek and other targets
DeepSeek, whose low-cost models gained international attention last year, launched a preview of its V4 model on Friday, optimized for Huawei chips. The company previously said its V3 model used data from web crawling, not synthetic data generated by OpenAI.
The cable also names Moonshot AI and MiniMax as targets of the campaign.
OpenAI warned U.S. lawmakers in February that DeepSeek was targeting the company and other leading American AI firms to replicate their models.
China's response
The Chinese Embassy in Washington rejected the accusations as "groundless" and called them "deliberate attacks on China's development and progress in the AI industry."
DeepSeek did not immediately respond to requests for comment.
Broader implications
The State Department push comes weeks before President Trump is scheduled to visit Beijing. The campaign could escalate tensions in the ongoing tech competition between the U.S. and China, potentially undoing a détente negotiated last October.
Multiple Western and some Asian governments have already restricted their institutions from using DeepSeek, citing data privacy concerns. Despite these bans, DeepSeek's models remain among the most widely used on international open-source AI platforms.
Your membership also unlocks:
