Regulation of AI in the Canadian workplace

The rapid pace of technology’s evolution has been extraordinary. Equally remarkable is the extent to which technology has changed how we live our lives given the efficiencies that computers, the Internet, and smartphones have brought to our workplaces and homes.

Computer scientists have been eager to give computers the power to simulate human intelligence by teaching them to recognize and analyze patterns in data so they can generate content and make decisions and recommendations. Today, artificial intelligence (AI) is a reality. Even at this early stage, the changes AI has brought to the world is of epic proportions.

As AI has the power to dramatically change our working lives, the purpose of this article is to summarize, at a very high level, the legal issues, risks and steps that employers should carefully consider in preparing for this technological revolution.

What are the risks of AI in the Canadian workplace?

Wrongful dismissal claims: If AI begins to perform the same duties as employees do now, this could reduce or even eliminate the need for employees to perform certain duties. When a Canadian employee’s duties are substantially changed, it could trigger a claim for damages on the basis they were constructively dismissed, causing significant employer liability. At common law, the employee’s damages may range from one to 24 months.

Privacy and confidentiality of information: AI systems may not adequately protect an employer’s confidential information, a third party’s confidential information in the employer’s hands, and personal information subject to privacy laws - the Personal Information Protection and Electronic Documents Act (PIPEDA) and the Personal Information Protection Act (PIPA) in Alberta and British Columbia.

Copyright infringement: The use of AI has the potential to raise copyright infringement issues. Is the employer entitled to input specific content into an AI system? Who owns the copyright in content created by AI? Will a third party’s copyright be infringed by the creation of AI generated content?

Discrimination in employment: Each province in Canada has human rights legislation. Prohibited grounds of discrimination include disability, race, ethnic origin, age, and similar grounds. Accordingly, human rights considerations may be raised when AI is used to make decisions related to recruitment, hiring, performance management, and pay equity. Employers using AI to make such decisions are vulnerable to claims that their decisions distinguish between individuals on grounds prohibited under human rights legislation.

Tort claims: When an individual is harmed by an AI system that has been developed or implemented by a business, the individual may claim the business is liable in negligence for the damage they experienced. Other Canadian tort claims may be made when AI systems create “deepfakes” that alter or duplicate a person’s voice or image – for example, interference with economic relations, intentional infliction of mental distress, and invasion of privacy for publicity that places the individual in a false light in the public eye.

New AI legislation

Canada has developed new legislation, including the Artificial Intelligence and Data Act (AIDA) - which is part of Bill C-27, the Digital Charter Implementation Act, 2022, Canada’s first attempt to create a national regulatory framework for AI that specifically focuses on the regulation of international and interprovincial trade and commerce in AI systems - and the Consumer Privacy Protection Act (CPPA), re-introduced to apply to those who develop and use “automated decision systems.”

Bill C-27 is not enacted or in force and, accordingly, neither is AIDA. AIDA is expected to come into force at least two years after Bill C-27 receives Royal Assent.

AIDA would be a national regulatory framework that would apply to private sector entities designing, developing, or deploying AI systems in trade and commerce – for example, to any “person” who, in the course of international or interprovincial commerce, carries out a “regulated activity”; is “responsible” for an AI System; or is “responsible” for a “high-impact system,” as such terms are defined in the statute.

When AIDA would become effective, the following obligations would be triggered for employers:

• Assess whether the AI System is a high-impact system.
• Establish “Mitigation Measures.”
• Publish a plain-language description of the system on a publicly available website.
• If there is a risk of material harm, notify the designated minister.
• Establish measures regarding anonymized data.

A person who violates these obligations or other specified offences would commit an offence under AIDA and be liable for significant fines.

The CPPA would apply to those who develop and use “automated decision systems,” as defined in the statute. They would be required:

• To make available, in plain language, “a general account of the organization’s use of any automated decision system to make predictions, recommendations or decisions about individuals that could have a significant impact on them.”
• If used in a way that could have a significant impact on individuals, to provide them with an explanation of the organization’s decision-making process upon request.

Voluntary Code of Conduct

In September 2023, the Voluntary Code of Conduct on the Responsible Development and Management of Advanced Generative AI Systems was launched. There are approximately 30 signatories who, prior to AIDA being proclaimed in force, committed to the responsible development and management of advanced generative AI systems in their operations, and to supporting “the ongoing development of a robust, responsible AI ecosystem in Canada.”

On March 21, 2024, Ontario’s Bill 149, Working for Workers Four Act, 2023 (Bill 149) received Royal Assent and only certain portions became effective. Bill 149 will, among other things, amend the province’s Employment Standards Act, 2000. One amendment is the addition of a new Part III.1, which will come into force on proclamation and require a statement disclosing the use of AI to screen, assess or select applicants for employment in publicly advertised job postings, subject to exceptions to be prescribed.

Quebec’s Act respecting the protection of personal information in the private sector (the Private Sector Act) contains a section regarding automated decision-making (s. 12.1), which came into force in September 2023. It requires any person carrying on an enterprise that uses personal information to render a decision based exclusively on an automated processing of that information, to inform the person concerned, whether customer or employee, no later than at the time it informs the person of the decision. The person concerned must also be informed, at their request, of the personal information used to render the decision; the reasons and principal factors and parameters that led to the decision; their right to have the personal information used to correct the decision; and be given the opportunity to submit observations to a member of the enterprise’s personnel who is in a position to review the decision. A monetary administrative penalty may be imposed on anyone who contravenes s. 12.1.

Bottom line for employers using AI

Employers designing, developing, or deploying AI systems in Canada should consider taking the steps required of them under AIDA and the CPPA when they would come into effect. This would demonstrate: their AI systems could be trusted from a consumer protection and human rights perspective; their AI research and innovation is conducted in a transparent and responsible manner; and their future compliance with AIDA and the CPPA when they become effective.

Employers who are not already signatories to the Voluntary Code of Conduct may consider becoming signatories to demonstrate that they value the model of a good corporate citizen.

Employers also should comply with any statutory requirements relating to AI that exist in the provinces in which they operate, if any.

To further mitigate risk, employers should address the legal risks of having AI in their workplaces by:

• Seeking the advice of experienced counsel in drafting employment agreements to permit the use of AI to perform any or all of the employee’s duties and responsibilities to mitigate the risk of constructive dismissal; and addressing their employees’ entitlement to reasonable notice or damages in lieu of reasonable notice upon the termination of an employee’s employment (including mass termination) due to a diminished need for employees caused by the deployment of AI in the workplace.
• Becoming knowledgeable about how data will be treated by any AI system they are considering using, and their contractual obligations regarding a third party’s confidential information.
• Developing and implementing appropriate workplace policies and entering into agreements with the developers of AI tools that ensure the confidentiality of a third party’s information, and their compliance with applicable privacy laws.
• Considering whether they are entitled to input specific content into an AI system in light of copyright infringement considerations.
• Ensuring the AI systems they use for employment-related decisions do not have flaws in their algorithms that would prevent them from being able to prove they made such decisions without bias and for objective, non-discriminatory reasons only, pursuant to applicable human rights legislation.

While the adoption and use of AI in the workplace is still in its infant stage in Canada, employers can take proactive action in preparation for these exciting developments.

Rhonda B. Levy is a knowledge management counsel for Littler in Canada, based in Toronto. Monty Verlint is a partner practicing labour and employment law at Littler in Toronto (www.littler.com).