DHS Uses AI to Craft Deportation Videos So Can the Public Still Trust Them

The US Department of Homeland Security has confirmed it is using artificial intelligence video tools from Google and Adobe to create and edit content shared with the public, a disclosure that sheds new light on how immigration messaging is being produced as deportation operations expand across US cities.

The revelation has fuelled fresh questions about transparency, verification, and how audiences can assess the authenticity of highly polished government videos appearing on social media.

DHS Confirms Use of AI Video Tools

A document released on Wednesday provides an internal inventory of commercial AI products used across US Department of Homeland Security for tasks ranging from document drafting to cybersecurity.

In a section covering public affairs materials, the agency confirms it is using Google's Veo 3 video generator through its Flow platform and Adobe Firefly to edit images and videos distributed publicly.

The document estimates DHS holds between 100 and 1,000 licences for these video generation tools. It marks the first time the agency has formally acknowledged the use of advanced AI video systems in content intended for public consumption.

How the AI Technology Works

Google's Flow combines the Veo 3 model with film-making tools that allow users to generate video clips and assemble full videos with sound, dialogue, and background noise. The result can be highly realistic footage that closely resembles conventional video production.

Adobe's Firefly, launched in 2023, can generate videos, images, soundtracks, and speech, with the company stating that it avoids using copyrighted material in its training data.

While the document confirms DHS access to these systems, it does not detail how frequently they are used or which specific public videos involve AI assistance.

Immigration Content Floods Social Media

Some videos have drawn attention for their cinematic style and hyper-realistic visuals, prompting speculation about AI involvement.

According to a report by KATV, Immigration and Customs Enforcement previously shared a video featuring an AI-generated Santa Claus encouraging undocumented immigrants to self-deport, highlighting how artificial intelligence has already been used in immigration-related messaging.

Until the release of the new DHS document, however, it was unclear which AI tools agencies were using to create such content.

Verification and Transparency Gaps

A key issue raised by the disclosure is verification. AI generated videos are difficult to independently identify, particularly once content is re-posted across platforms. Adobe offers watermarking options to signal AI generated media, but those markers do not always remain intact after uploading and sharing.

DHS has not indicated that it labels or discloses when AI tools are used in specific videos. As a result, members of the public have no clear way to determine whether a piece of immigration messaging was edited or created with AI, or produced using traditional methods.

Tech Worker Pushback and Company Silence

The use of AI by immigration agencies has also prompted internal criticism within the tech sector. More than 140 current and former Google employees and over 30 from Adobe have recently urged their companies to take a public stance on collaboration with immigration enforcement agencies.

Neither Google nor Adobe has issued a response to the DHS disclosure. DHS itself has also not commented beyond the released documentation.

Wider AI Use Across DHS

Beyond video production, the document shows DHS relying on AI tools for drafting reports, summarising long documents, supporting coding tasks, and operating facial recognition systems. The use of AI is presented as part of a broader modernisation effort rather than a programme limited to communications.

What remains unclear is how DHS intends to address public understanding and oversight of AI assisted messaging. With immigration content spreading rapidly online and the line between traditional and AI generated media increasingly blurred, the disclosure has opened a new debate over trust, verification, and transparency in government communication.