US Government Website Briefly Used Chinese AI Tool Flagged by FBI

Why Did a Federal Site Test a Tool the FBI Warned Against?

The Federal Register website incorporated an open source Chinese artificial intelligence search model for a short period in September 2026, despite the FBI having previously labeled the technology as malicious. The tool was integrated into the site’s search function to improve document retrieval for public notices and regulatory filings. Officials confirmed the integration was temporary and removed after internal security reviews raised concerns about data privacy and potential foreign influence. The incident highlighted tensions between adopting innovative AI tools and maintaining cybersecurity safeguards within federal systems.

The AI model in question, developed by a Chinese research group, had been flagged by the FBI in an earlier advisory as posing risks due to its potential for data harvesting and unauthorized access patterns. Despite these warnings, the Federal Register’s technical team tested the tool to evaluate its efficiency in handling large volumes of legal text. The trial lasted only a few days before being discontinued. Agency spokespeople stated that no breach occurred and that all user data remained secure during the test period. The decision to use the tool sparked debate among cybersecurity experts about the balance between innovation and risk in government digital infrastructure.

What Are the Implications for Future AI Use in Government?

The Federal Register sought to modernize its search capabilities to keep pace with increasing public demand for access to government documents. Traditional search functions struggled with complex queries involving regulatory language and cross-referenced statutes. The Chinese AI model demonstrated strong performance in semantic understanding and contextual matching during internal benchmarks. Officials said the test was conducted in a sandbox environment with no connection to live databases. However, the FBI’s prior warning about the model’s behavior in real-world scenarios led to heightened scrutiny. The episode underscored the challenges agencies face when evaluating foreign-developed AI tools without clear federal guidelines on vetting such technologies.

Following the incident, the General Services Administration announced a review of procedures for evaluating third-party AI tools used on federal websites. The goal is to establish clearer criteria for assessing security risks, especially those involving foreign-developed software. Lawmakers have called for greater transparency in how agencies select and deploy AI systems, particularly when they involve international contributors. The Federal Register has since reverted to its previous search system while exploring domestically developed alternatives. Experts suggest the event may accelerate efforts to create a federal AI approval process akin to the FedRAMP program for cloud services.

Was any sensitive data exposed during the test of the Chinese AI tool? No, officials confirmed that the trial was conducted in an isolated environment and no breach or data leakage occurred.

Frequently Asked Questions

Why did the FBI label the AI model as malicious? The FBI cited concerns over the model’s potential for data exfiltration and behavior consistent with unauthorized access attempts in prior assessments.

Will the Federal Register consider using similar AI tools in the future? The agency said it remains open to AI-enhanced search but will apply stricter vetting standards, preferably using tools developed within allied nations or under federal oversight.