VicOne to stress test robot safety at DEF CON 34

VicOne will bring a Physical AI Safety Stress Test to DEF CON 34 in Las Vegas on Aug. 6-9, 2026, using a digital twin CTF to show how cyber risk can change what AI-driven robots perceive, decide and do. The challenge is designed to help robotics makers and operators find cyber-to-safety risks before deployment.

Why it matters: - Robotics safety is expanding beyond motors and mechanics to include cyber risk, AI models, cloud services and communications stacks. - A cyber issue can change robot behavior even when hardware is functioning normally. - Manufacturers and operators need better ways to test how attacks could affect safety before robots are deployed near people, equipment and critical assets.

What happened: - VicOne said it will bring the Physical AI Safety Stress Test to the DEF CON 34 Robotic Hacking Community. - The challenge will run Aug. 6-9, 2026, at the Las Vegas Convention Center in Las Vegas. - The event will be hosted in West Hall 4, Level 1, Room 1309. - The hands-on capture the flag challenge is limited to 80 teams. - The company positioned the test as a digital-twin exercise for physical AI security scenarios.

The details: - The challenge is built to show how cyber risk can affect what AI-driven robots perceive, decide and do before those systems operate in real environments. - VicOne said its LAB R7 has tracked more than 20 public Physical AI security incidents and vulnerability disclosures over the past 18 months. - The challenge uses six mission areas: prompt injection targeting robot AI reasoning, poisoned AI policy models, adversarial visual inputs, cloud and API security weaknesses, robotics communication protocol weaknesses including ROS 2 and DDS, and embedded firmware trust-chain weaknesses. - Each mission is designed to show how a cyber condition can influence robot perception, decision-making, task execution, operational boundaries or trust assumptions. - The test runs in a Real2Sim digital twin built in a physics-based robotics simulator. - The environment models the Reachy Mini Lite and LeKiwi mobile manipulator, both of which will also appear in the RHC show-floor environment. - Each team gets an isolated environment through the challenge API, so participants can explore attack paths without exposing live robots to unsafe conditions. - VicOne framed the format as a virtual escape room for AI robots. - VicOne said the exercise is meant to help robotics makers and operators identify where safety validation must expand as robots become more autonomous, connected and AI-driven.

Between the lines: - The message is that robot safety cases now need cyber-to-physical testing, not just mechanical validation. - VicOne is extending its automotive cybersecurity work into robotics, using the challenge to translate vulnerability research into a safety-testing model. - The focus on prompt injection, poisoned models and protocol weaknesses suggests the company sees AI reasoning, connected services and trust chains as emerging weak points. - The emphasis on a controlled digital twin points to a practical route for testing dangerous scenarios without putting real hardware or people at risk.

What's next: - Robotics makers and operators can use the event to evaluate cyber-to-safety risks before deployment and during pre-release validation. - The challenge may also inform post-market monitoring, incident response and site-safety planning for autonomous systems. - VicOne said it will continue applying its automotive cybersecurity and cyber-physical risk validation expertise to Physical AI systems.

The bottom line: - VicOne is using DEF CON 34 to argue that robot cybersecurity is now a safety issue, not just an IT issue.