An activist began running photographs of masked ICE officers through PimEyes — a commercial facial recognition search engine — to identify agents at enforcement operations. Cross-referencing the PimEyes hits with public records produced names and home addresses. Senator Blackburn pressed PimEyes's CEO in November 2025. No criminal charges have been filed.
What happened
In August 2025, [Politico reported](https://www.politico.com/news/2025/08/29/ai-unmasking-ice-officers-00519478) that an activist had begun using PimEyes — a commercial facial recognition search engine that lets anyone upload a photo and find other photos of the same face on the open web — to identify Immigration and Customs Enforcement officers from images taken at enforcement operations. PimEyes returned hits even when officers were masked, by matching unmasked features visible on social media or in older photos. The activist then cross-referenced the names returned by PimEyes against public-records and broker sites to get home addresses. Senator Marsha Blackburn sent a [follow-up letter to PimEyes CEO Giorgi Gobronidze in November 2025](https://www.blackburn.senate.gov/2025/11/technology/blackburn-presses-facial-recognition-company-for-evading-responsibility-for-activists-using-its-technology-to-dox-ice-agents) pressing the company on its role in enabling doxxing of federal agents. PimEyes responded that it complies with GDPR and other applicable laws. As of April 2026, no criminal charges have been filed against the activist.
What happened
In August 2025, Politico published an investigation describing how an activist had been using PimEyes — a commercial facial recognition search engine — to identify ICE officers from photos and video taken at enforcement operations.
PimEyes is an open-web product. Anyone can upload a photo of a face and get back other photos of the same person scraped from the public internet. In this use case, the activist uploaded images of masked ICE officers. PimEyes matched those faces to unmasked photos elsewhere on the web — social media accounts, older news photos, public profile pictures — and returned identities. The activist then ran those names through public records and people-search broker pages to get current home addresses.
Senator Marsha Blackburn sent a follow-up letter to PimEyes CEO Giorgi Gobronidze in November 2025 pressing the company on its role in enabling doxxing of federal agents. PimEyes responded that it complies with GDPR and applicable laws. No criminal charges have been filed against the activist as of April 2026.
How it started
Two consumer-grade tools, stacked. The first is facial recognition that works against the open web. The second is a people-search broker stack that resolves a name into a home address.
The mask doesn't help once unmasked photos exist somewhere. A wedding photo from a decade ago, a department softball team picture, a photo of you with a relative on a public profile — any of those will give PimEyes enough to match. Once the face matches a name, broker pages do the rest.
This is a pure off-the-shelf chain. No insider source. No leak. No hack.
Why this case matters
The PimEyes chain converts an operational moment — being on scene at an enforcement action — into a doxxing pipeline that runs in minutes. The technology exists. It's marketed and sold to consumers. There is no legal mechanism that takes the face-match step off the table.
The leverage point is the second step. Once the activist has the name, the home-address lookup runs through people-search broker pages. That stack is regulable, and removable.
What this means for you
If you're an ICE agent, Border Patrol officer, or any sworn federal officer whose face is visible in any public-facing context, the PimEyes chain assumes you can be identified. The defensive question is what an attacker can do once they have your name.
Federal LE is not covered by the Lieu Act, which covers judges. State Daniel's Law analogs vary on whether they cover federal officers. The thing that closes the second step regardless of jurisdiction is broker removal. We file opt-outs across 200+ sites and re-run them when the listings come back.
For more on facial-recognition exposure and the broker stack that completes the chain, see /threats/social-media-osint and /data-brokers.
Editorial rules: Only public, already-reported incidents. Never name a non-public victim. Always end with the prevention takeaway tied to our service. Cite at minimum one public source per claim.
What would have prevented this
PimEyes is the new front end. The back end is the same broker stack that has been resolving names into home addresses for a decade. Wearing a mask at an operation no longer prevents identification, because facial recognition can match against unmasked photos posted years earlier. Once the name is known, the lookup that produces the home address runs through the same people-search pages that any first responder is exposed to. Federal LE has no equivalent of the [Lieu Act](/laws/lieu-act), which covers judges. State analogs to [Daniel's Law](/laws/daniels-law) vary on whether they cover federal officers. The defensive layer that always works is removing your name and address from the broker pages that the second step relies on. We do that across 200+ sites and re-check every two weeks.
Public sources
- AI is unmasking ICE officers — Politico, 2025-08-29
- Blackburn Presses Facial Recognition Company For Evading Responsibility For Activists Using Its Technology To Dox ICE Agents — U.S. Senate (Office of Sen. Blackburn), 2025-11-01