Video Protection and In-Car Cameras: Who Has to Prove What?

In the previous article, we saw that the "I have nothing to hide" argument reverses the burden of justification: it is not up to each person to prove their innocence, but up to whoever wants to restrict a freedom to justify the request. The same reversal appears, in other forms, in two systems that have become ordinary.

Reassuring words

Camera surveillance is officially called "video protection" in France. Cameras pointed at the driver are called "driver assistance". These words describe a benefit, not a technique. A camera that watches a face watches a face, whatever name it is given.

What artificial intelligence actually does

In so-called "algorithmic" video protection, artificial intelligence analyzes images continuously, with nobody in front of the screens. Technically, programs detect and classify people and objects, follow their movements, count them and flag anything that matches predefined situations: an abandoned object, a dense crowd, a person on the ground, someone crossing a prohibited area, a fire. Facial recognition is excluded by law.

The system was trialed for the 2024 Olympic Games, then extended, and new texts plan to extend and widen it further. Reports on the trial mention false alerts and limited reliability for some detections. What remains is the essential point: the criteria for what counts as "abnormal" are set by software whose workings are not public.

A camera pointed at the driver

Since 2024 for new models, and for all new vehicles from July 2026, European regulations require systems that detect driver drowsiness and distraction. The text requires a function, not a specific component, but in practice these systems rely on a camera, often infrared, pointed at the face: gaze direction, head position, eyelids.

The driver did not choose it and cannot refuse it. What the manufacturer keeps, where it is sent, for how long: these questions have no answer that anyone can check for themselves.

Measurements taken from the body

The law reserves the word "biometric" for data used to identify a person. This choice of vocabulary lets these measurements fall outside the most protected category. But in the plain sense of the words, measuring the face, gaze or behavior of a living person is measuring a living being.

Data of this kind can make it possible to identify someone. What is collected today to detect fatigue could be re-analyzed tomorrow with other programs. You cannot change your face the way you change a password: once captured, the exposure is permanent.

The point is not to claim that this is being done. It is to note that nothing guarantees it will not be, and that the person concerned has no way of knowing.

Who has to prove what?

The ethical questions of the previous article come back:

• Purpose. What exactly is the system for, beyond the general goal announced?
• Limit. What will it never do, and who can check?
• Control. Who decides what is "abnormal", and who watches the watcher?
• Consent. When a system is imposed, what is left of the person's choice?

It is not up to citizens to demonstrate that an abuse will occur. It is up to those who impose these systems to guarantee, verifiably, that it will not. As with everything else: we do not have to prove that we deserve our freedom.