The Department of Homeland Security’s Customs and Border Protection (CBP) agency has procured software from Fivecast, an AI company specializing in detecting “sentiment and emotion” in online content, according to documents obtained by 404 Media.
CBP employs this technology to analyze open-source information related to travelers who may pose a threat to public safety, national security, or lawful trade and travel.
Fivecast’s software, which also offers AI-enabled object recognition and risk term detection, boasts the ability to collect targeted data from major social platforms like Facebook, Reddit, and smaller communities such as 4chan and 8kun. Despite its capabilities, concerns about accuracy and utility are raised, as CBP deploys multiple AI-powered systems in its monitoring efforts, including U.S. citizens.
Researchers note that ransomware hackers are moving at an accelerated pace, with the median dwell time for ransomware incidents dropping from nine to five days in the first half of 2023. Improved endpoint detection could be pressuring hackers to expedite attacks, although executing complex double-extortion attacks in under five days remains unlikely. These hackers favor deploying their final payload outside regular business hours, particularly on weekends, to maximize successful strikes when staffing levels are low.
CBP’s use of AI continues to raise questions about privacy and accuracy, especially given its previous utilization of AI tools linking social media posts to individuals’ Social Security numbers. The documents reveal that CBP has invested millions in Fivecast’s software, signaling the agency’s growing reliance on AI technologies for monitoring purposes.
Despite CBP’s commitment to privacy and civil rights, concerns remain about the potential misuse of these tools, their accuracy, and the scope of surveillance they enable.
