ICE Is Paying $15M for AI to Protect Agents From Doxing, but Its Own Officials Say: 'I Don't Trust It'

US Immigration and Customs Enforcement is launching a new initiative designed to help protect its personnel from doxxing, but some agency officials are questioning whether they can trust the technology behind it. A confidential internal memo signed by Acting ICE Director David Venturella and leaked to The Intercept by ICE employees announced the agency's 'Doxing Mitigation Initiative,' which involves cybersecurity company ZeroFox and is intended to help safeguard ICE personnel and their families.

The initiative comes alongside a reported $15M Department of Homeland Security contract with ZeroFox for software licenses supporting ICE operations and investigations.

'It Sounds Like Inside Surveillance'

ZeroFox markets technology designed to identify exposed personal information, monitor online threats, and facilitate the removal of certain exposed data. The precise capabilities being deployed under ICE's new initiative, however, remain unclear.

The lack of clarity has prompted concern inside the agency. 'It sounds like inside surveillance,' one ICE official told The Intercept anonymously, citing fears of retaliation. The official questioned whether technology meant to protect personnel from exposure could also be used to identify agency critics or whistleblowers.

'If things go sideways, they may just throw all your info out there, make you out to be the problem or dox you,' the official said. 'I don't trust it.' The official was expressing a concern about potential misuse, not providing evidence that ICE intends to use the technology for that purpose.

Another ICE official also raised separate privacy concerns, saying some personnel already turn to third-party services to reduce their online exposure. 'I frankly don't trust the feds with that,' the official said. 'I'm not really keen on giving my PII (personally identifiable information) to a company that doesn't have a solid track record with government contracts.'

ZeroFox Has Faced Scrutiny Before

ZeroFox has previously faced scrutiny over its social media monitoring technology. In 2015, its 'Crisis Management Report' identified Black Lives Matter organisers DeRay McKesson and Johnetta Elzie as 'high' severity 'physical' threats during protests in Baltimore following the death of Freddie Gray after he was injured in police custody. The report did not identify specific evidence that McKesson or Elzie posed a physical threat. ZeroFox co-founder Evan Blair later said the designation was not intended to suggest that the two organisers meant to cause harm and that the system likely identified them because of their large Twitter followings.

The FBI also switched its social media alerting provider from Dataminr to ZeroFox effective 1 January 2021. A Senate investigation later found that the contract transition impeded the FBI's ability to detect threats on social media in the lead-up to the 6 January attack on the US Capitol. Internal FBI emails also showed concerns about the new system's operability, including that automated searches relied upon by Washington Field Office personnel had not yet been set up after the ZeroFox contract took effect.

Expert Warns of Potential 'Honey Pot' Situation

Richard Forno, associate director of the cybersecurity institute at the University of Maryland, Baltimore County, said that the technology could plausibly be used to identify whistleblowers. 'It's plausible that this tech could be used to seek out whistleblowers,' he said. He described a hypothetical 'honey pot situation' in which officials could deliberately seed information, see whether it was passed to a journalist, and use that trail to identify a potential leaker. There is no evidence presented that ICE is currently using ZeroFox this way.

Sophia Cope, an attorney with the Electronic Frontier Foundation, said that protecting people from the publication of personal information can be a legitimate safety concern. 'However,' she said, 'if ICE is instituting this program to shield officers and the agency itself from public accountability, especially when information is already publicly available, then that's a serious concern and flies in the face of our democratic values.'