The US Department of Homeland Security has been quietly using AI video generators from tech giants Google and Adobe to create and edit public-facing content, according to a newly released government document.
The disclosure arrives at a particularly charged moment, as immigration agencies flood social media with posts supporting President Trump’s mass deportation efforts, some of which appears to have been made with artificial intelligence.
A document made public on Wednesday offers the first detailed look at DHS’s arsenal of commercial AI tools, spanning everything from document drafting to cybersecurity management.
DHS AI Inventory Reveals Use of Google Veo and Adobe Firefly for Content Creation
Among the most notable revelations: the agency is using Google’s Veo 3 video generator and Adobe Firefly, with an estimated 100 to 1,000 licences for these powerful content creation tools.
The inventory also shows DHS relies on Microsoft Copilot Chat for drafting documents and summarising lengthy reports, along with Poolside software for coding tasks and various other commercial AI products. Google, Adobe, and DHS did not immediately respond to requests for comment about the disclosure.
These findings shed new light on how agencies like Immigration and Customs Enforcement, a division of DHS, might be producing the enormous volume of content they’ve been sharing on X and other platforms as immigration enforcement ramps up across American cities. ICE’s recent social media activity has included posts celebrating “Christmas after mass deportations”, references to Bible verses and Christ’s birth, images showing the faces of arrested individuals, and recruitment ads targeting potential agents.
The agencies have also repeatedly used music in their videos without obtaining proper permissions from artists.
The Commercial AI Tools Powering DHS Content Machines
While some videos posted by these agencies have long carried the telltale signs of AI generation, there’s been no concrete evidence until now about which AI models might be powering this content machine.
The newly released document provides that missing puzzle piece, confirming for the first time that DHS is indeed using commercial AI generators to create material shared with the American public.
That said, verifying whether any specific piece of content was made with AI, or identifying which company’s tools created it, remains nearly impossible. Adobe does offer watermarking options to disclose when videos are AI-generated, but these markers often get stripped away when content moves across different platforms and social media sites.
The document reveals that DHS specifically uses Flow, a Google tool that combines the Veo 3 video generator with a complete suite of filmmaking capabilities. Users can generate video clips and assemble entire productions with AI assistance, including sound effects, dialogue, and background noise that can make the results strikingly realistic.
Adobe launched its Firefly generator in 2023 with promises that it wouldn’t use copyrighted content in either training or output. Like Google’s offerings, Adobe’s tools can generate videos, images, soundtracks, and speech. However, the document doesn’t provide details about exactly how DHS is deploying these video generation capabilities.
The revelations come as employees at major tech companies push their employers to take a stand. More than 140 current and former Google workers and over 30 from Adobe have been pressuring company leadership in recent weeks to speak out against ICE operations and the shooting of Alex Pretti on January 24.
DHS and ICE Expand Use of Facial Recognition and Content Creation Tools
So far, Google’s leadership has remained silent on these demands. Last October, both Google and Apple removed apps from their stores that were designed to track ICE sightings, citing safety concerns.
A second document released Wednesday provided additional insights into DHS’s use of more specialised AI products, including a facial recognition app that ICE has been deploying, which was first reported by 404Media back in June.
As AI-generated content becomes increasingly sophisticated and harder to distinguish from human-created material, questions about transparency and accountability in government communications take on new urgency. The public now knows these agencies have access to some of the most advanced content creation tools available—but the full extent of how they’re being used remains largely in the shadows.
