DHS orders federal agencies to remove DeepSeek software within 30 days

The Department of Homeland Security announced Monday that all products from Chinese artificial‑intelligence startup DeepSeek must be removed from federal agency IT systems and that the company is barred from future government contracts, the order said. The directive, first issued in July and now published in the Federal Register, cites “multiple threat vectors” in DeepSeek’s large‑language models and requires agencies to purge the software within 30 days.

The move comes as Washington tightens scrutiny of Chinese technology amid rising geopolitical tensions and concerns that AI tools could be weaponized for espionage or data theft. In recent months, the administration has tightened rules on Chinese‑made hardware, banned TikTok on government devices, and warned about supply‑chain risks in AI. Lawmakers have warned that AI models can embed hidden code that transmits user data to servers abroad, a risk officials say DeepSeek’s systems have not adequately mitigated.

Industry analysts note that DeepSeek, founded in 2022, quickly gained attention for its open‑source models that rival those of OpenAI and Google. The company has secured funding from Chinese venture capital firms and operates data centers in mainland China, raising alarms about potential government access under China’s national‑security laws. A senior DHS official said the agency’s review found the models could exfiltrate sensitive information, generate disinformation, or be repurposed for cyber‑attacks, prompting the ban. The decision aligns with similar actions taken against other Chinese AI firms, such as Baidu and iFlytek, which have faced restrictions in the United States.

Federal agencies now face the logistical task of auditing codebases, replacing DeepSeek services with approved alternatives, and documenting compliance for oversight bodies. DHS warned that failure to comply could result in penalties for individual departments. DeepSeek has not publicly responded, but a spokesperson for the China‑based firm previously argued that its technology complies with international security standards. The ban may spur further legislative proposals to tighten AI procurement rules, and it underscores the growing friction between U.S. security policy and the global AI marketplace.