ONVIF’s Leo Levit on how AI and standards are reinventing trust in surveillance

As generative AI transforms the world of surveillance — from smarter video analytics to the alarming rise of deepfakes — the need for digital trust has never been greater. In this exclusive interview, Leo Levit, chairman of ONVIF, discusses how the security industry is responding to these disruptive changes, the role of international standards in preserving the authenticity of video evidence, and why collaboration across stakeholders is critical to navigating this evolving landscape.

Since its establishment in 2008, ONVIF has played a pivotal role in creating global standards for the interoperability of IP-based physical security products. With more than 500 members worldwide and over 25,000 conformant products, the organisation continues to set the benchmark for trusted and secure video surveillance systems. Now, as the world grapples with AI-driven manipulation and increasingly complex surveillance networks, ONVIF’s mission has never been more vital.

1. How is the rise of generative AI reshaping the global security and surveillance industry, and what key trends are emerging in response to the technology’s growing influence?

The integration of generative AI into the security and surveillance industry is a double-edged sword. On one hand, it enhances capabilities, enabling more efficient video analytics, such as behaviour detection, crowd formation, or object tracking, as well as predictive surveillance geared towards crime prevention. On the other hand, it introduces new risks, for example in the form of manipulated video, fake identities, and misinformation. The rapid development of AI-generated content is pushing the industry to evolve frameworks that can quickly distinguish between authentic and altered footage.

One of the most notable trends is the increasing demand for video authentication and digital watermarking to verify the integrity of footage and its production source. This is especially important for evidence management.

Another trend is greater collaboration between camera manufacturers, software providers, and standards bodies to develop safeguards that work across systems and regions. AI is not only reshaping how surveillance is conducted—it is changing our trust in video authenticity, a cornerstone of the security industry.

2. With deepfake technology becoming more accessible, how are businesses and governments adjusting their security frameworks to protect against digital manipulation and fraud?

Governments and organisations are increasingly prioritising video verification methods in their security protocols. This includes implementing camera tampering detection, encrypting video files, and integrating secure-by-design principles to protect footage from the moment of capture to the moment it’s viewed—whether by police, the courts, or private entities.

At ONVIF, we are developing a method called media signing, which proves that a video has not been altered since it left the camera sensor that captured it. Securing the video at its source is key to ensuring its authenticity.

In forward-thinking regions like the UAE and the wider GCC — where smart cities and AI surveillance are strategic priorities—deepfakes are increasingly seen as a national security issue.

Governments are evolving frameworks that mandate authenticity, auditability, and transparency. The Dubai Police’s move toward zone-based security exemplifies how surveillance is becoming more proactive and technologically sophisticated. But as more data enters the system, ensuring that the footage is real becomes as important as detecting threats.

3. What role do you see international standards organisations like ONVIF playing in ensuring digital trust and interoperability in an increasingly AI-driven surveillance ecosystem?

We play a critical role in safeguarding the integrity and interoperability of security systems in an era of rapid AI evolution. As a global and open industry forum, ONVIF was created to standardise communication between IP-based physical security products, enabling devices and software from different manufacturers to work together seamlessly. Today, ONVIF has become the de facto standard for interoperability in surveillance, with our products used by governments, businesses, and institutions worldwide.

As generative AI becomes more embedded in video and analytics workflows, maintaining trust in the output of surveillance systems is more important than ever. Our standardised interfaces support this by enabling secure and consistent data exchange across platforms—an essential feature when building systems that rely on AI to detect and respond to events.

Moreover, ONVIF’s work on standardising video authentication tools — such as media signing — will allow law enforcement and legal professionals to verify whether video footage has been tampered with, regardless of the device manufacturer. It’s not just about compatibility — it’s about building trust into every layer of the surveillance and security infrastructure.

4. As public scepticism around digital evidence increases, what innovations or best practices are emerging to help restore trust in video as a reliable source for legal, corporate, and security decisions?

That’s a great question. Distrust in video evidence can lead to concerns about reasonable doubt in legal and corporate settings. If AI manipulation undermines the current legal precedents around video admissibility, courts may be forced to redefine what qualifies as valid evidence. This could mean excluding footage where authentication can’t be proven.

Restoring trust hinges on two pillars: transparency and traceability. Innovations such as encrypted digital signatures and embedded forensic watermarking are gaining traction across public and private sectors. These tools help verify when, where, and how a video was captured — and crucially — whether it has been tampered with.