Our daily online activities produce a massive amount of data that reveals intimate details about our lives. Knowingly or unknowingly, what we do online can reveal whom we voted for, what we buy, who our friends are, where we work, and even if we still want to work there. A slew of private companies now systematically collect, aggregate, and analyze this data on an unprecedented scale, using it to create and test new, invasive surveillance tools to monitor what the American public is up to. Agencies across the federal government—and local law enforcement across the nation—are eagerly entering into expensive contracts to obtain these tools, which they appear to be using for purposes ranging from tracking weather events to tracking protesters. Because the proliferation of these new public-private surveillance partnerships threatens to undermine our freedom of speech and association online, as well as our privacy, today the Knight Institute filed two Freedom of Information Act requests to inform the public about the scope and the details of these partnerships.

­The social media surveillance market is crowded with companies selling an array of tools. Clearview AI, for example, says it has aggregated over ten billion images from social media platforms and other websites, and used them to develop facial recognition software. This software purportedly enables users to identify a person based on a single photograph and provides links to other publicly available photographs of the person on the internet—including on social media websites, which often carry intimate details about those individuals. And then there’s Dataminr, which sells tools and services that enable location tracking and keyword searches for the purported purpose of keeping government actors apprised of breaking events in real time. Its First Alert service is designed in particular to track the initial stages of emergencies, including natural disasters and the spread of COVID-19. Other major surveillance technology providers include Babel Street, ZeroFox, Geofeedia, Talkwalker, and Meltwater.

The social media surveillance market is equally crowded with federal agencies seeking to use those tools in a variety of contexts. Reports on the nationwide protests following the murder of George Floyd in 2020, for example, described federal law enforcement agencies’ use of social media surveillance tools such as Dataminr to target protesters, and no fewer than six federal agencies conducted investigations with the help of facial-recognition software. More recently, the FBI and other agencies conducted social media surveillance following the January 6, 2021, attack on the U.S. Capitol, with at least three agencies using facial-recognition software to support criminal investigations. Other agencies—including the Fish and Wildlife Service, the Department of the Treasury, and the Department of Commerce—also have sought to or currently use these surveillance tools, whether for one-time searches or to conduct ongoing surveillance of particular users or keywords. Government procurement requests suggest that various branches of the military, such as the Navy, also appear to be interested in social media monitoring technology to track coverage and commentary of their actions. While the use of such monitoring technology varies across agencies, the technology is often used for cybersecurity, rapid news alerting, communications and “brand management,” threat intelligence, and law enforcement investigations of specific individuals or groups.

The FBI, for example, uses social media surveillance tools for investigative purposes, as it acknowledged following the January 6, 2021, attack on the U.S. Capitol. Yet the nature and extent of this surveillance is unclear. Statements by the FBI director suggest that its use of social media surveillance tools is limited, but internal documents indicate otherwise. What is clear is that the FBI has contracted with Dataminr since at least 2016 for its Twitter surveillance capabilities. Additionally, the FBI has recently contracted with ZeroFox after seeking a tool capable of “constant monitoring of publicly available data on social media platforms,” and notifications based on keywords, location, or subject matter, among other things.    

Similarly, in January 2020, the Criminal Investigation Division of the U.S. Army (“CID”) began soliciting private companies for a tool with the ability to “search against all social media that allows public access, such as Twitter, Facebook, Instagram,” and others, while also providing keyword searching, location-based searching, and face-matching capabilities. CID chose Clearview and contracted with the firm for multiple licenses. While available data suggest that CID has run over 1,300 searches with Clearview, the purposes or guidelines surrounding these searches remain unknown.    

Despite the increasingly widespread use of these new surveillance tools, many have questioned their reliability, as well as their constitutionality. In particular, researchers have raised concerns that facial-recognition technology is biased against women and people of color. And even if improved accuracy reduces the risk of false identification, the government’s use of such tools still threatens privacy and free expression.  Social media contains massive amounts of sensitive information about individuals’ beliefs and associations, as the earlier example regarding the 2020 social justice protests demonstrates. The collection, analysis, and marketing of this data by private companies to the government are likely to chill First Amendment rights, especially when so little is known about the nature and extent of such practices. 

That’s why we are submitting FOIA requests to CID and the FBI today. CID has contracted for Clearview AI and acknowledged its use publicly, but how and for what purpose remain unclear. Similarly, the precise nature of the services for which the FBI contracts with Dataminr and ZeroFox is still undisclosed. Learning the scope of the services provided under these contracts is necessary to inform the public of potential burdens to their First Amendment rights and to address concerns of bias built into the tools the government uses for social media surveillance.