The FBI is exploring artificial intelligence for its national-security threat-screening operations, with a March 2026 procurement document seeking technology capable of analysing large government datasets and using predictive modelling to identify patterns and locate additional relevant information.
The FBI AI plans form part of an effort to automate some of the manual work carried out by its Threat Screening Center.
The request for information, obtained by Reason, does not explicitly say the proposed system would predict whether an individual will commit terrorism or identify political dissidents before they act.
Instead, the document calls for a capability that can analyse existing records for similarities, patterns and correlations and predict where additional relevant information could be found across interconnected government systems.
FBI's AI Plans Include Predictive Modelling
The FBI's March 2026 request sought information from potential technology vendors about AI enhancements for the Threat Screening Center, or TSC. The centre is responsible for maintaining and supporting the US government's consolidated terrorism watchlist and related screening operations.
The Trump administration is creating an AI system for "pre-crime policing," hoping to leverage government data on private citizens to predict crimes before they happen. pic.twitter.com/eY370bzFkz
— FactPost (@factpostnews) July 29, 2026
Among the requirements listed in the document is 'Predictive Modeling Using Enhanced Data with Traceable Lineage'. The proposed capability would examine existing records for similarity, pattern alignment, and attribute correlation, while helping identify where additional relevant information could be obtained across federated systems.
The FBI AI system described in the procurement documents would therefore be designed to help investigators and analysts work through large amounts of information more efficiently. The document does not state that the system would predict that a particular person will commit a future crime.
Reason interpreted the predictive-modelling language as potentially allowing the TSC to use AI to help predict who might be a terrorist. However, that is an interpretation of the procurement language rather than an explicit statement of the FBI's intended operational use.
The request also indicates that the FBI wants greater automation around data enrichment and other time-intensive tasks. The emphasis on traceable data lineage is significant because it suggests the agency is seeking systems that can show where information came from and how it was used in the analysis.
The procurement process also calls for attention to AI governance and oversight, including requirements connected to the FBI's AI Ethics Council.
Threat Screening Mission Grows
The procurement comes after the FBI expanded the remit of the organisation responsible for federal terrorism watchlisting.
In March 2025, the FBI announced that the Terrorist Screening Center had been renamed the Threat Screening Center. The agency said the new name reflected an expanded mission covering terrorism and other national-security threats, including transnational organised crime.
The change means the centre's work is no longer described solely through the lens of terrorism screening. However, the March 2026 AI procurement does not establish that the proposed technology is intended to identify people based on political beliefs or protected activities.
The FBI's own description of the watchlisting system also places limits on how people can be added. The agency says individuals cannot be placed on the terrorism watchlist because of race, ethnicity, religion, beliefs, or activities protected by the First Amendment, nor can people be nominated simply on the basis of guesses or hunches.
The scale and reach of the watchlisting system nonetheless make the proposed use of AI significant.
A January 2026 Government Accountability Office report found that state and local law-enforcement officers can query individuals encountered during routine interactions, including traffic stops, against the terrorist watchlist.
A separate 2021 data exposure involved more than 1.9 million records believed to have been drawn from the federal terrorist watchlist. That figure should not be treated as the current number of people on the list because records can include different entries, aliases, and other information.
Pushback Against Watchlist System
The legal history of the watchlisting system also shows why its use has attracted scrutiny. In FBI v. Fikre, the Supreme Court considered a challenge involving a man's placement on the No Fly List. In Tanzin v. Tanvir, the court considered claims brought by Muslim plaintiffs who alleged they had been placed on the No Fly List in retaliation for refusing to become government informants.
Neither case established that the federal watchlist itself is unconstitutional.
The FBI has also publicly discussed expanding its use of artificial intelligence across national-security operations.
FBI Director Kash Patel has said the bureau is working with major technology companies to strengthen its technology infrastructure and expand the use of AI in counterterrorism.
The March 2026 procurement, however, does not establish that the FBI has already deployed an AI system capable of predicting individual terrorist activity. It shows that the Threat Screening Center was seeking information from vendors about technology that could automate data analysis, correlate information across systems, and provide predictive modelling capabilities.
The precise operational role such technology would eventually play remains undefined in the public procurement documents.
