Are We Creating Video or Evidence?

For as long as the security industry has existed, we have sold a simple promise: the recording is the truth. A camera captures what happened and the footage proves it. Our clients pay for that promise and investigators, compliance teams, auditors, claims adjusters, law enforcement and the courts all rely on it. That promise is now being tested in ways most of us are unprepared for. Artificial intelligence-based deepfake tools have shifted from novelty to commodity and hyper-realistic deepfakes are now widespread. Research by Kawping found that by late 2025, approximately one in five videos shown to a new YouTube account were AI-generated. As public awareness grows, trust in video content is declining, creating a trust gap and raising questions about authenticity. Industries related to security are experiencing a significant rise in AI-manipulated videos. While these synthetic counterfeits do not originate from our sector, this trend will directly affect us. Consider this scenario: a clear surveillance recording becomes central to a dispute. While no one disputes its content, the opposing party questions its authenticity. The footage is not dismissed as fake but is set aside because its authenticity cannot be verified. This shift changes what integrators, dealers, installers, and channel partners must provide. While customers will continue to value cameras, resolution, storage and security, they will increasingly require systems that can prove the authenticity of digital evidence. This is already happening. In court cases, people are now claiming that real recordings could be AI-generated, a tactic called the deepfake defense. The federal judiciary’s committee on evidence rules has spent the last two years working on rules for AI-generated and AI-altered media. Video no longer gets the benefit of the doubt, and now the burden of proof is on whoever recorded and stored the footage. For us, that means our customers and our own teams. The Video Evidence Fix Has to Come From Us Let’s be honest. Our approach to chain of custody hasn’t changed in decades. We use encryption, watermarking, localized hashing, and provenance metadata to protect recordings, and these tools have served us well. But these methods were made for a time when the main threats were intercepted files and simple edits, not advanced AI manipulation and attacks on trust. They also have a bigger flaw: each one is tied to the recording it protects. A watermark can be removed. A hash stored in the footage is only as reliable as the system that keeps it. Metadata can be changed along with the file. Anything that stays inside the evidence can be changed there. As AI transforms how video is evaluated, the security industry must adapt to remain relevant. Our customers will increasingly ask a critical question and expect us to provide the answer: can you prove this video is authentic? As video authenticity becomes harder to assume, integrators should be thinking about how the systems they deploy can support independent verification. Customers may not use that language today, but they will increasingly need a way to confirm that footage is real, unchanged and tied to a reliable record of how it was created, stored, exported and shared. Simply put, we need a system that can check if video recordings are real without relying only on the hardware or software that made them and that can’t be fooled by artificial intelligence. The protection must also be effective. While the instinct may be to counter AI with AI and detect fakes, this approach leads to an unwinnable arms race. Researchers at UC San Diego demonstrated that deepfake detectors could be deceived more than 80 percent of the time, even by attackers unfamiliar with the detection methods. The tech has advanced further, with each new detector informing the next generation of attacks. More Than a Technology Discussion Detection will remain part of the conversation, but integrators should also consider how authenticity is established from the beginning. The stronger long-term approach is to help customers preserve confidence in legitimate recordings as early as possible, then maintain that confidence as video moves through storage, export, transfer, review, and potential legal use. That confidence also needs to survive the way video is actually used. In many projects, footage does not stay in one system. It may be exported from a VMS, shared with law enforcement, uploaded to a case management platform, sent to an insurer, or reviewed by legal teams weeks or months later. Integrators should understand how each handoff affects the evidentiary value of the footage and help customers document the process clearly. This is not just a technology discussion. A 2025 law enforcement futures study, written by a 25-year police veteran for California’s POST Command College, concluded that agencies will soon need to fingerprint digital evidence as soon as it is collected and keep it in tamper-evident, independently verifiable records, precisely to prove in court that nothing changed after capture. This issue does not require us to become cryptographers. Rather, it requires us to recognize that authenticity is now a core offering, not an add-on. For integrators and channel partners, this is an obligation and an opportunity. Customer discussions will shift from cameras, resolution, and storage to a more valuable proposition: proof that a system’s recordings reflect actual events. For integrators, this creates an opportunity to deliver value beyond hardware alone. The most prepared partners will be able to ask better questions of their vendors, document deployment decisions more carefully, train customers on proper evidence handling, and help ensure that recorded video can stand up to scrutiny when it matters most. Our role has never been solely to record video; it has been to create evidence. If we do not adapt, AI will reduce our recordings to mere video when our customers most need evidence. We owe them more. We owe them what we promised. Jason Crawforth is founder and CEO of SWEAR. The post Are We Creating Video or Evidence? appeared first on Security Sales & Integration.