SecEdge launches AI-PASSPORT to verify AI model trust

SecEdge unveiled AI-PASSPORT at Embedded World North America 2026 to give AI models cryptographic identity, provenance and integrity checks. The company is demonstrating the system on NVIDIA Jetson Thor with NVIDIA Nemotron open models as concern grows over poisoned or impersonated AI models.

Why it matters: - SecEdge is targeting a growing AI supply-chain risk: models can be modified, poisoned or impersonated and then redistributed as if they were authentic. - AI-PASSPORT is meant to help enterprises, robotics teams and edge deployments verify that a model is the approved version and has not been tampered with. - SecEdge is positioning model trust as a safety and compliance issue, not just a technical feature, as AI moves from the data center to edge devices.

What happened: - SecEdge unveiled AI-PASSPORT at Embedded World North America 2026 in Anaheim, California. - The company is demonstrating the product live at booth 6226 and in the NVIDIA Robotics Lab. - The demo runs on NVIDIA Jetson Thor using open models from the NVIDIA Nemotron family. - SecEdge also plans technical sessions focused on AI trust across enterprise, robotics and edge deployments.

The details: - AI-PASSPORT is a cryptographically verifiable trust credential for open-source and proprietary AI models. - The system establishes a model’s identity, provenance and integrity. - The credential travels with the model. - Verification does not require exposing or inspecting the model’s contents. - Developers, deployers and auditors can use the system to confirm the model came from the expected source and has not changed since signing. - SecEdge says the approach helps organizations deploy only trusted original models. - SecEdge is part of a full-stack functional safety architecture for Physical AI. - NVIDIA Halos AI Systems Inspection Lab is described as the world’s first ANAB-accredited inspection program for AI and functional safety. - The lab is intended to help partners validate system integration and accelerate third-party certification.

Between the lines: - The launch signals a push to apply familiar software integrity concepts, like signed provenance and verification, to AI models. - That matters because model tampering is harder to spot than traditional code compromise and can affect downstream systems without obvious warning signs. - SecEdge is tying AI-PASSPORT to a broader safety and certification story, which could appeal to regulated industries and hardware partners. - CEO Sami Nassar said the need to know which model is running, where it came from and whether it has been tampered with is becoming a safety and compliance requirement.

What's next: - SecEdge is hosting a 15-minute Tech Talk in the NVIDIA Robotics Lab titled “Whose AI Are You Running?” - The session is scheduled for Wednesday, Sept. 23, from 1 to 1:15 p.m. - The talk will focus on medical AI applications. - SecEdge leadership will be available afterward to speak with attendees. - The company is directing readers to more information and a contact email at info@secedge.com.

The bottom line: - SecEdge is betting that AI model provenance will become a core requirement for safe deployment at the edge.