More and more reports are coming out detailing US government agencies’ desire to expand and “refine” their surveillance operations by including detection of sentiment and emotion, by using “AI”-powered software.
One of them is Customs and Border Protection (CBP), which is part of the Department of Homeland Security (DHS), and which, according to internal documents obtained through freedom of information act (FOIA) requests, hired a third party in order to scan online posts for “risk terms and phrases.”
The company that provides this tool, Fivecast, does so in multiple languages. Fivecast also sells surveillance of video and image material through what it calls AI-enabled object recognition.
The company’s marketing itself as being able to scrape data for these purposes not only from large platforms such as Facebook and Reddit, but also 4chan, 8kun, and Gab.
It apparently doesn’t stop there, because there is also anti-trafficking and anti-propaganda capability baked in there somewhere, at least according to what are said to be leaked statements made by one employee.
404 Media reports about all these findings stemming mostly from a number of FOIA documents it has had access to, adding that CBP responded to the outlet’s queries by stating that data used in the process is “open source” (meaning, publicly available online).
Those targeted by the software, meanwhile, are travelers either arriving to or leaving the US who are considered a potential threat to a fairly wide range of interests: public and national security, trade and travel. And these “persons of interest” can be foreign and US citizens alike.
But while enough is known about what the tools of this kind – and CBP is said to be deploying a number of them – are supposed to be able to achieve, just how accurate and useful they actually are remains something of an enigma.
