AI is reshaping how the world creates and consumes content. Users can now leverage large language models (LLMs) to draft emails, compile reports, or summarize documents – and those same capabilities are being applied to video. Tools capable of creating high-quality videos with little to no effort, or altering real videos to look drastically different, are widely available. Anyone who has scrolled social media or YouTube recently has likely asked themselves: is this video real or fake? As with any technology, AI is being used not just by well-meaning individuals, but also by criminals with ill intent.

These capabilities have created significant challenges for the security industry. How can we definitively say that a security video has not been altered using today's advanced AI tools? This has emboldened cybercriminals and cast doubt on video evidence, creating an environment where it is easier than ever to raise questions about the authenticity of digital content. It is a challenge device manufacturers cannot afford to ignore. Today's customers need mechanisms in place to keep their devices secure and clearly validate the authenticity of their video.

How Attackers Are Leveraging AI

The number of incidents involving manipulated or AI-generated content continues to grow, and the impact is surprisingly far-reaching. Public figures, organizations, and brands have all been affected by synthetic images, video, and audio that spread rapidly online before their authenticity is verified. While these incidents are hardly limited to the security industry, they underscore the broader challenge: increasing skepticism about digital content. For organizations that rely on video as evidence, maintaining confidence in the authenticity and integrity of recordings has become more important than ever.

When people assume anything they see on video could potentially be fake, it erodes trust across society. That may sound abstract, but it has a tangible impact on industries like security. Courts, law enforcement agencies, and other stakeholders are increasingly scrutinizing the authenticity and provenance of digital evidence, meaning those submitting video must be prepared to prove its authenticity and demonstrate that the chain of custody was maintained from the device itself all the way to the courtroom. As generative AI tools make it easier to create convincing synthetic content, organizations need reliable methods to clearly demonstrate the integrity of video evidence.

The Challenge with Traditional Video Systems

Rebuilding trust in video is not something that can be accomplished by any single party, but security providers – and device manufacturers in particular – play an important role. It will require a significant shift from the way video surveillance has typically worked. Historically, video collected from surveillance devices was streamed to a network video recorder (NVR) and sent to a video management system (VMS), where it could be exported and used as needed. That video could be digitally signed from the VMS, but there was no reliable way to authenticate whether the video itself was legitimate or confirm the specific device it was sourced from.

This created clear chain of custody issues. A video submitted as evidence in a legal case might be reviewed by multiple lawyers and their offices. To confirm that a clip had not been manipulated, you would need to compare it directly to the original video. Even then, this could only confirm that the video matched what was exported by the VMS – it could not confirm that the video matched what was originally recorded by the device itself. It is a subtle but important difference. Simply put, if you want your video to be trusted, you must be able to prove its authenticity before it leaves the device.

Device IDs Play a Critical Role

Signed video has emerged as a key method of addressing this issue. Device IDs provide cryptographically verifiable device identities that support secure onboarding, authentication, and trust establishment throughout a device's lifecycle. Today, those same IDs can also be used to positively identify the device that produced a given video. Before the video leaves the camera, cryptographic signatures can be embedded in the video stream using a device-unique signing key. This enables verification that the footage originated from a specific device and has not been altered after capture – making it possible to determine if a single frame has been changed. It also provides a significant volume of information, including the model of the device and the firmware it is running.

Even with complete verification, there are ways to manipulate video – but they leave critical evidence behind. A bad actor might use an AI tool to spoof an element of the video after the fact, but doing so strips signed video of its metadata. When the video is cryptographically verified, any modification to the footage invalidates the signature, making tampering evident. While it is technically possible to create signed video within the VMS – previously known as watermarking – this is not considered a valid form of verification, as the video could have been altered before reaching the VMS. The only way to guarantee the authenticity of a video is to sign it at the source, on the device itself.

A Holistic Approach to Ensuring Authenticity

Device IDs and signed video are among the most important ways to confirm integrity and authenticity, but they represent just one element of a broader approach. Practices like Secure Boot and encrypted communications play an important role as well, and customers should look to work with partners that provide a full suite of video security and authenticity mechanisms.

Trust starts at the device level. Secure Boot helps ensure that a device starts only with authorized software, establishing a chain of trust from hardware through operating system and application execution. Encryption of data in transit and at rest helps protect video and device communications from unauthorized access and tampering. Signed video represents the final step in the chain, providing a method to verify that recorded video has not been altered after capture. As trust becomes increasingly important in a world of growing cyberthreats and misinformation, these capabilities should be considered standard – not optional.

"Trust but verify" has long been an unofficial mantra within the security industry. In an era where synthetic media is becoming increasingly sophisticated, trust in video cannot rely on perception alone. By combining secure device identity, trusted hardware, cryptographic verification, and signed video, the security industry can help ensure that video remains a reliable source of evidence and insight. The goal is not simply to trust video, but to provide clearly verifiable proof of authenticity.