Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Autonomous systems security protects machines that sense, decide, and act in the physical world, where a manipulated input is not a wrong answer but a vehicle in the wrong lane.

What Is Autonomous Systems Security?
Autonomous systems perceive their surroundings, decide what to do, and act on that decision without a person in the loop. Self-driving vehicles are the familiar example, but the category also covers delivery drones, warehouse robots, agricultural machinery, maritime vessels, and the increasingly automated control loops in industrial plants. Securing them means accepting a hard constraint that does not apply to information systems: when the decision is wrong, something moves.
Each stage fails in its own way, and the failures compound.
| Stage | Attack | Consequence |
|---|---|---|
| Sense | GNSS spoofing, LiDAR and radar injection, camera dazzling, adversarial markings | The system builds a false picture of the world |
| Decide | Model manipulation, poisoned training or map data, planner edge cases | A correct picture produces a dangerous action |
| Act | CAN bus or actuator command injection, firmware tampering | The decision is overridden entirely |
| Update | Compromised over-the-air channel or build pipeline | A fleet is affected at once |
GNSS spoofing is the most mature. Incidents of mass position falsification affecting aircraft and shipping have been reported around conflict zones since 2023, and civilian receivers without authentication accept the strongest signal available. Sensor injection against automotive LiDAR and radar has been shown repeatedly in laboratory conditions, creating phantom obstacles or hiding real ones. On the decision side, Tencent’s Keen Security Lab showed in 2019 that small markings on a road surface could steer a lane-keeping system across a lane boundary, a practical instance of the adversarial example problem. On the action side, the 2015 remote compromise of a Jeep Cherokee over its cellular link remains the reference case for why in-vehicle networks need segmentation, and the same reasoning applies to a drone whose flight controller trusts any message on the bus.
Automotive is the most developed. UNECE Regulation No. 155 requires a certified cybersecurity management system covering the whole vehicle lifecycle, and Regulation No. 156 covers software update management; both apply to new vehicle types from July 2022 and to all new vehicles produced from July 2024 in the UK and EU. ISO/SAE 21434 is the engineering standard manufacturers use to demonstrate conformity. The Automated Vehicles Act 2024, which received Royal Assent on 20 May 2024, establishes the UK framework for self-driving vehicles, including authorisation requirements and an in-use regulatory scheme, with cybersecurity integral to the safety case. Drone operations sit under Civil Aviation Authority rules, and industrial autonomy falls within the NIS Regulations where it supports essential services, overlapping with operational technology security. NCSC principles for connected places and for operational technology are useful even outside those regimes.
Assurance here cannot stop at the software. A credible programme combines code and protocol testing, hardware and bus-level work, radio frequency assessment of positioning and telemetry links, and adversarial testing of the perception models, then validates the safety layer by trying to defeat it. Simulation covers scale; bench and track testing covers the physics. At Privia Security, we assess embedded controllers, radio interfaces, and control networks through our Hardware Penetration Test Services and OT-SCADA Penetration Test.
IoT devices mostly sense and report; autonomous systems decide and act, often at speed and with mass behind them. That difference changes the priorities: availability and integrity outrank confidentiality, safety engineering and security engineering have to produce one combined argument, and a control that fails open is unacceptable because the fallback is a physical action rather than a missing data point.
Not entirely on civilian frequencies, but it can be detected and contained. Multi-constellation receivers, inertial dead reckoning, signal-quality monitoring, antenna arrays that check direction of arrival, and authenticated services such as Galileo OSNMA all raise the bar. The design goal is that losing trust in positioning triggers a defined safe behaviour rather than silent navigation on false data.
Yes. The UK applies UNECE R155 and R156 for vehicle type approval, with R155 requiring a cybersecurity management system across development, production, and post-production. Manufacturers typically demonstrate conformity through ISO/SAE 21434 processes, and suppliers are drawn in because the obligation covers the whole supply chain rather than just the vehicle assembler.
Enforce deterministic limits that no learned component can override: maximum speed for the detected context, geofence boundaries, minimum separation, torque and braking limits, and a transition to a defined safe state when inputs are inconsistent or absent. Implement it separately from the model, in code that can be reviewed and formally reasoned about, and test it by deliberately feeding the system bad perception.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.