When Faith Meets the Algorithm: Can Employees Opt Out of AI at Work?

Pope Leo XIV’s Magnifica Humanitas is less an anti‑technology statement and more a reminder of something employers sometimes overlook: the workplace is not a values‑free zone. 

As organizations move quickly to embed generative AI into drafting, hiring, customer service, and decision‑making, a new and underappreciated risk is emerging. 

Employees are not just questioning whether AI is useful. Some are starting to ask whether using it conflicts with their religious beliefs. 

For employers, the legal risk is not the request itself. It is how the request is handled. Treating AI objections as preferences rather than potential human rights claims is exactly the kind of misstep that leads to liability. 

Given that AI mandates are being rolled out faster than policies are being developed, this gap is where disputes are likely to arise. 

How This Will Show Up in the Workplace 

AI‑related accommodation issues are not theoretical. They are likely to arise in practical, everyday situations, such as: 

• A professional is required to use AI to generate first drafts but refuses, believing that delegating judgment to a machine conflicts with their religious understanding of authorship and responsibility. 

• An employee is expected to input work product or client data into an AI system, but objects based on beliefs about stewardship, control, or moral accountability for how information is used. 

• A manager is asked to rely on algorithmic recommendations for hiring or performance evaluation but resists using them in decision-making based on sincerely held beliefs about fairness or human judgment. 

Each of these situations raises the same core question: is this a preference, or is it a bona fide creed-based objection? 

The answer to that question will determine whether a duty to accommodate is triggered. 

Not Every Objection Triggers a Legal Duty 

Ontario’s human rights framework provides a clear starting point. In Public Health Sudbury & Districts v Ontario Nurses’ Association, 2022 CanLII 48440 (ON LA), an arbitration arising from a COVID-19 vaccination policy, the employer distinguished between protected grounds (i.e. creed) and unprotected objections (i.e. personal belief or philosophy). 

That distinction carries directly into the AI context. An employee who expresses general resistance to AI is likely expressing a preference, not a protected ground. 

Religious Objections Must Be Taken Seriously 

Canadian law does not require religious belief to be universally accepted or formally endorsed. The focus is on sincerity and personal conviction. 

In the decision above, the arbitrator relied on a key principle from the Supreme Court of Canada’s Amselem case: a person’s religious belief does not have to be officially recognized or required by their faith to be protected. 

In practical terms, there are two questions to ask. 

1. Does the employee have a sincere religious belief that guides how they think they should act? 

2. Does the workplace requirement interfere with that belief? 

If the answer to both is yes, the employer needs to consider whether accommodation is possible and if it would create undue hardship. 

Importantly, courts and arbitrators are not there to decide whether someone’s interpretation of their faith is “right” or “wrong.” The focus is on whether the belief is genuine and whether the workplace rule creates a real conflict. 

AI Will Surface New Types of Conflicts 

As AI becomes embedded in workflows, employees may object to AI-driven drafting, contributing data to intelligent systems, or relying on algorithmic decision-making. These issues will become operational realities as AI use expands. 

The Real Question: Is AI Use Essential? 

Accommodation often turns on whether the requirement is truly integral to the role. 

This is where many employers will face scrutiny. If AI is framed internally as “best practice” or “expected,” but employees can still perform their work effectively without it, then it becomes harder to argue that opting out creates undue hardship. 

By contrast, where AI is embedded in workflow, required for efficiency, or tied to client expectations, the analysis shifts. The more essential the tool, the stronger the employer’s position. 

A Structured Approach for Employers 

• Define the AI requirement clearly 

• Understand the employee’s specific objection 

• Assess possible alternatives 

• Document any undue hardship analysis 

Where Employers Could Get into Trouble 

• Treating objections as preferences without assessment 

• Applying blanket AI mandates 

• Failing to document decisions 

• Inconsistent handling across teams 

The Takeaway

AI adoption is moving quickly. Workplace rights analysis is not. 

For employers, this is not simply a question of compliance. It is a question of risk management and operational clarity. 

Before these issues surface, leadership teams should be asking: 

• Where is AI use actually essential to performance? 

• Where is it a preference or efficiency tool? 

• Are managers equipped to distinguish between a complaint and a protected claim? 

Organizations that cannot answer these questions consistently will struggle to defend their decisions. 

The opportunity is to get ahead of the issue. Build AI policies that define when use is required, where flexibility exists, and how objections will be assessed. Train managers to respond in a structured way. Document decisions carefully. 

AI is not just changing how work gets done. It is reshaping how workplace rights will be tested. 

Employers who treat it that way will be better positioned to move quickly without creating unnecessary legal exposure. 

For more information, contact George Hamzo at Lerners LLP.