AI-Enabled Monitoring Systems In Vehicles — How Far Can Employers Go?

Artificial intelligence offers innovative solutions to strengthen workplace safety. However, as noted in a recent Tribunal decision in Québec,¹ its use must be consistent with fundamental rights and privacy obligations.

Background

In 2023, a bus transportation company ("Employer") implemented a monitoring system ("System") primarily powered by artificial intelligence ("AI").

The system featured two cameras: one monitoring the road and the other, inside the cabin, monitoring the driver from head to knees. Video footage was recorded continuously from when the engine was turned on until 15 minutes after it was shut down. The footage was automatically stored for 90 days on the supplier's servers, and the Employer could view or retrieve the contents at its discretion.

This System also had the following capabilities:

• ability to view live video feed (up to ten minutes per month per bus);
• discretionary access to recorded videos, even in the absence of a detected incident;
• automatic reports of risky behaviour (using cell phone, running stop signs, failing to wear a seatbelt, etc.);
• using recorded videos for training purposes, sometimes without the driver's consent;
• displaying a photo of the driver assigned to each bus in service on the vehicle location page, with the image, captured by the camera inside the cab, automatically updated every two minutes; and
• the Employer's ability to copy and process a 2-10 second video clip of an incident captured by the System for potential training purposes, sometimes without the driver's consent.

The Grievance

The Union filed a grievance challenging certain features of the System on the grounds that they violated the right to dignity, privacy and fair and reasonable working conditions under sections 4, 5 and 46 of the Charter of Human Rights and Freedoms² ("Charter").

More specifically, the Union objected to continuous video surveillance via a camera aimed at the driver, even when the service was not being provided; the storage of recordings for purposes other than those justifying their collection; real-time monitoring of drivers; and the use of their image without consent.

Decision

The Tribunal recognized that the primary function of the System — the detection of safety incidents — was justified and necessary. In particular, it argued that the installation of the System for the purposes of protecting the safety of drivers and other road users was justified, especially since the Employer's drivers operate heavy vehicles.

However, the Tribunal found that the following features of the System were excessive and infringed the drivers' fundamental rights protected under the Charter:

• viewing real-time images captured by the camera inside the cabin;
• storing video data from the camera monitoring inside the cabin that was unrelated to reported security incidents, with discretionary access granted to certain individuals designated by the Employer;
• publishing the driver's photo on the vehicle location page; and
• internally sharing videos of incidents involving identifiable drivers, without their consent.

After a detailed analysis of the situation, the Tribunal found that the Employer had not demonstrated a rational connection between these practices and the goal of preventing incidents involving the safety of persons, nor that they were proportionate or justified on other reasonable and legitimate grounds.

Accordingly, the grievance was allowed in part, and the Employer was ordered to cease the practices infringing the Charterand to pay each affected driver a compensatory indemnity of $100 for moral injury resulting from the violation of their right to privacy, dignity and fair and reasonable working conditions.

Takeaways

The use of any AI-enabled system must comply with Charter rights — such as the right to privacy, dignity and fair and reasonable working conditions — as well as legal obligations under privacy protection laws.³

In particular, privacy laws, which have quasi-constitutional status, provide for various obligations regarding the use of artificial intelligence in systems, such as:

• conducting privacy impact assessments prior to the deployment of an AI system in order to identify and minimize privacy risks;
• conducting an analysis of the need to implement an AI system — the system must not only be useful, but also address a real and objective problem;
• obtaining the individual's consent (e.g., through a consent form) to use and share images captured by the AI system for specific purposes;
• establishing procedures for accessing videos captured by the AI system;
• setting and respecting retention periods for videos captured by the AI system.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.