 |
| DeepSeek Artificial Intelligence |
Foreign Ministry spokesperson Mao Ning stated on Tuesday that Beijing “has consistently opposed the overstretching of the concept of national security and the politicisation of issues related to science, technology, and trade.” She added that China would “resolutely safeguard the legitimate rights and interests of Chinese companies.”
The comments came one day after DHS publicly detailed an order—originally signed in late July by Homeland Security Secretary Markwayne Mullin—requiring civilian executive-branch agencies to remove DeepSeek products and any third-party services built using the company’s models from their information technology systems. Agencies are also barred from acquiring such products going forward. The notice, published in the Federal Register, invokes the Federal Acquisition Supply Chain Security Act of 2018 and follows a recommendation from the Federal Acquisition Security Council.
Scope and Stated Rationale of the Ban
The restriction applies to federal civilian agencies, including the Departments of the Treasury, State, and Justice. It does not cover the Department of Defense or the intelligence community, which operate under separate authorities and had already imposed their own limits on DeepSeek earlier. The order explicitly extends beyond DeepSeek’s own applications and APIs to “third-party products and services developed using DeepSeek implementations,” closing off potential workarounds in which another vendor simply wraps the same underlying models.
DHS described DeepSeek’s models as presenting “multiple threat vectors.” Key concerns include the risk that prompts and data submitted by U.S. government employees or contractors could transfer sensitive information to servers located in China. Chinese national security laws grant the government extensive authority to access private company infrastructure. Officials warned that such data, combined with publicly available information, “could produce actionable intelligence on U.S. persons and/or government agencies.” Additional risks cited include the possibility of malware or other security vulnerabilities in the models themselves, as well as practices such as collecting personally identifiable information and digital fingerprinting of users.
Several U.S. agencies and states had already restricted DeepSeek on government devices prior to the DHS order. The Commerce Department, the Navy, NASA, and the House of Representatives took earlier steps, as did states including New York, Tennessee, and Virginia. The July 24 order formalized a broader exclusion from civilian procurement channels.
DeepSeek’s Rapid Rise and the Broader AI Contest
DeepSeek, based in Hangzhou, emerged as a disruptive force in early 2025 when it released high-performing open-weight models at a fraction of the cost associated with leading U.S. systems. Its R1 reasoning model challenged assumptions about American technological primacy and contributed to significant market volatility among semiconductor and AI-related stocks. The company has continued iterating aggressively. It released the V4 series in preview form in April 2026, featuring models with million-token context lengths and mixture-of-experts architectures designed for efficiency.
Subsequent updates, including a V4.1 Flash variant released in September and further refined this month, have emphasized faster task completion, stronger agentic capabilities, and competitive benchmark performance in coding, reasoning, and multimodal tasks while remaining far less expensive to run than many proprietary Western alternatives.
The U.S. government and industry have repeatedly accused Chinese AI firms, including DeepSeek, of using techniques such as model distillation—systematically extracting capabilities from frontier American models—to accelerate their own development. In September 2026, the National Security Agency, FBI, and Cybersecurity and Infrastructure Security Agency jointly warned that several Chinese companies had conducted such efforts “at an industrial scale” since at least late 2024, sometimes routing requests through multiple pathways to evade terms of service. Beijing has rejected these claims, insisting that Chinese AI progress stems from independent innovation and self-reliance.
Strategic Context and Implications
The ban fits a pattern of supply-chain security measures targeting Chinese technology providers in sensitive domains. Similar logic underpinned earlier restrictions on Huawei telecommunications equipment and certain TikTok uses on government devices. For federal contractors and software vendors, the order creates practical compliance burdens: any tool that routes queries to DeepSeek models, even indirectly or via model routers, may now fall under the prohibition. Agencies have limited windows to seek exceptions.
From Beijing’s perspective, the action continues a trend of U.S. measures that constrain Chinese access to advanced chips, manufacturing equipment, and markets while portraying Chinese firms as inherent security risks. Chinese officials have consistently argued that such policies hinder global scientific collaboration and amount to protectionism under a security guise.
Despite the rivalry, the two countries agreed in September to establish a communication channel for AI-related incidents, and the issue is expected to surface again when leaders meet.
Analysts note that the restrictions highlight a fundamental tension in the AI era. Open-weight models like DeepSeek’s lower barriers to advanced capabilities and accelerate innovation worldwide, yet they also complicate traditional notions of technological control and data sovereignty. U.S. officials prioritize preventing potential intelligence collection and ensuring the integrity of government systems. Chinese counterparts emphasize reciprocity and the right of their companies to compete on performance and cost.
DeepSeek itself has not issued a detailed public response to the latest order. The company’s trajectory—from relatively obscure research lab to a major player whose models power applications globally—illustrates how quickly the competitive landscape can shift when efficiency and open release strategies challenge established leaders. Whether the DHS action meaningfully limits DeepSeek’s broader commercial adoption outside government channels remains an open question. Private-sector use in the United States and internationally continues, and open-weight releases allow local deployment that may reduce some data-transfer risks while still raising questions about model provenance and potential embedded vulnerabilities.
As AI systems become more deeply embedded in critical infrastructure, decision-making tools, and everyday software, governments on both sides are treating model selection as a matter of strategic security rather than purely commercial choice. The DeepSeek ban is the most comprehensive U.S. federal action against a single Chinese AI lab to date. It is unlikely to be the last such measure in an intensifying technological contest.
Samuel O. Sulaimon is the founder and Editor-in-Chief of Naijapiz. He covers Global, News, Sports, Finance, Faiths, Education, Politics and Technology.
Post a Comment