L'ICE paie une société d'intelligence artificielle controversée pour cacher l'identité des agents

U.S. Immigration and Customs Enforcement is deploying artificial intelligence tools developed by a controversial firm to obscure the identities of its agents from the public. According to an internal email sent to all agency employees, the agency is rolling out a new “Doxing Mitigation Initiative” designed to protect ICE personnel from being exposed through online data breaches. The memo, signed by Acting ICE Director David Venturella, was obtained by The Intercept and reveals that ZeroFox, an AI-focused security company, is providing the technology for this effort. The document describes ZeroFox as offering “modernized protective technology solutions” aimed at safeguarding ICE employees and their families. The agency’s parent body, the Department of Homeland Security, has already awarded ZeroFox a $15 million contract in July for software licenses intended to aid in ICE operations and investigations. Public procurement records show that ZeroFox will deliver services to the Office of Intelligence within ICE’s Homeland Security Investigations unit. The firm specializes in AI-driven platforms designed to detect and monitor potential data leaks across social media and the dark web. Its website states that it continuously tracks and maps users' digital footprints, claiming to “never stop finding, mapping, and monitoring your digital presence.” Despite these disclosures, details on how the new initiative will operate remain unclear. Neither ICE nor ZeroFox has responded to inquiries regarding the specifics of the program. An anonymous ICE official, speaking to The Intercept, described the arrangement as resembling “inside surveillance,” expressing concerns over the potential misuse of such technology. ZeroFox has previously been involved in identifying individuals associated with political activism. During the 2015 protests following the death of Freddie Gray in Baltimore, the firm reportedly flagged Black Lives Matter organizers as threats. Similarly, after the January 6, 2021, attack on the U.S. Capitol, the FBI encountered challenges in tracking online extremists using ZeroFox’s tools, with some agents giving the platform a derogatory moniker. Richard Forno, associate director of the cybersecurity institute at the University of Maryland, Baltimore County, noted that it is unsurprising for ICE to collaborate with firms that may have problematic pasts. He warned that such technology could be leveraged to identify whistleblowers or critics of former President Donald Trump. Forno suggested that if ICE seeded information into the public domain, it might inadvertently expose sources or create situations akin to “honey pots” to trace leaks. In 2015, ZeroFox provided its services to Baltimore authorities without charge, delivering a list of “threat actors” identified via its social media monitoring systems. These included peaceful Black Lives Matter activists, who were categorized as physical threats due to their offline activities. Co-founders of ZeroFox later admitted that the classifications were influenced by follower counts and the distinction between online and real-world engagement. Sophia Cope, a senior staff attorney at the Electronic Frontier Foundation, acknowledged that limiting the exposure of personal information online is a reasonable desire. However, she cautioned against the broader implications of ICE implementing such measures to shield itself from public scrutiny. She emphasized the importance of transparency and accountability in government actions involving advanced technologies.