Government Use Of Artificial Intelligence Expands To Social Media

One of the few things that Republicans and Democrats agree on is that the use of artificial intelligence by the United States government should be limited. A recent article from 404 Media based on Freedom of Information Act requests shows that at least one of the tools being explored by the Department of Homeland Security uses AI-generative analytics to monitor social media. While the tool is intended to search for terrorists, the capabilities make it likely that the tool might be used inappropriately to find the identities of people who have yet to commit a crime.

The FOIA documents show that the U.S. Customs and Border Patrol, an agency within DHS, has spent millions on software designed to predict a person or group’s thoughts and emotions. The software was developed by Fivecast which claims the ONYX program can filter through popular social media platforms and bulk data to create a “Person of Interest.” The company claims the tool is designed “specifically for that purpose” in documents obtained by 404 Media.

Beyond simply relaying messages and images, which the tool can also do, it has the ability to use AI to determine an individual’s sentiments and emotions over time. In a document released by 404 Media, the tool used a 2019 New York Times article to generate a POI of an individual associated with the group, Boogaloo. The software identified an individual, tracked his data and location, then provided images, phrases, and even a map of emotions that showed the individual targeted by the tool expressed heightened anger and frustration in early 2020.

CBP claims that the use of AI is an essential part of the agency’s mission to stop human and drug trafficking, identify terrorists both domestic and foreign, and is only used for legitimate purposes within the scope of the law.

Fivecast claims the tool could also be useful for anti-propaganda missions, a touchy subject with both Republicans and Democrats. A June 2023 article by Morning Consult shows that more than half of members of both parties fear AI will be used for political campaigns and should be regulated.

In 2021, Yahoo News reported on the use of AI technology by the Counter Network Division of the CBP that gathered data on U.S. journalists. Though an investigation was opened by the Office of the Inspector General, no charges were filed. In part, the use of AI may fall beyond the scope of existing law, a fact acknowledged by a CNP employee who claimed that the division is “the ones making guidelines” in the use of AI to catch criminals.