ONVIF Media Signing Add-on targets AI-generated video manipulation in surveillance

The release of this candidate version addresses the growing threat of AI-generated deepfakes in legal and insurance proceedings by establishing a chain of custody at the hardware level. For the streaming infrastructure market, this signals a shift toward mandatory provenance metadata, as interoperability between different camera brands and verification tools becomes essential for evidentiary standards. As generative AI tools become more sophisticated, the industry must move beyond simple watermarking toward these robust cryptographic methods. Watch for the final specification release in late 2026 to see if major IP camera manufacturers move to claim conformance across their existing product profiles.

ONVIF's Media Signing Add-on arrives amid a broader industry push to establish cryptographic provenance for video content across multiple standards bodies and technology stacks. The Coalition for Content Provenance and Authenticity (C2PA), backed by Adobe, Microsoft, Intel, and the BBC, has been developing its own content credentials specification for media provenance since 2021, and its 2.1 specification was adopted by major camera manufacturers including Sony, Nikon, and Leica for embedding tamper-evident metadata at capture time. While C2PA focuses on general media authenticity across photography, journalism, and entertainment, ONVIF's effort specifically targets the physical security vertical where evidentiary chain-of-custody requirements are legally binding. The two frameworks are complementary rather than competitive, with C2PA handling broad media provenance and ONVIF addressing surveillance-specific interoperability across multi-vendor camera deployments.

The regulatory environment is accelerating demand for verifiable video provenance. The European Union's AI Act, which entered into force in August 2024 and begins phased enforcement through 2026, requires that AI-generated or manipulated content be clearly labeled and detectable, with penalties reaching up to 7% of global annual turnover for non-compliance. In the United States, several states have introduced deepfake legislation targeting synthetic media used in legal proceedings and political contexts. For the surveillance industry specifically, courts in multiple jurisdictions have begun questioning the admissibility of video evidence that lacks verifiable integrity metadata. ONVIF's working group, which includes members from Axis Communications, Bosch, and Hanwha Vision, is positioning the Media Signing Add-on as the technical answer to these emerging legal requirements, giving security integrators a standardized path to demonstrate evidentiary compliance regardless of camera manufacturer.

From a technical standpoint, the ONVIF approach differs from watermarking solutions by embedding cryptographic signatures directly into the video stream at the encoder level rather than overlaying visible or invisible markers. This aligns with broader trends in streaming infrastructure where content integrity verification is becoming a network-layer concern. Ericsson's network architecture team has described how distributed AI inference at the edge will require deterministic infrastructure with built-in authenticity verification for sensor and video data, a vision that extends beyond telecom into physical security deployments. The ONVIF specification leverages existing Profile M metadata streaming capabilities, meaning cameras already conformant with ONVIF Profile M can potentially implement media signing through firmware updates without hardware replacement. This backward-compatibility strategy mirrors how the streaming industry has handled codec transitions, where software-upgradable infrastructure reduces adoption friction for operators managing large installed bases of devices.