AI rules in Canada
The AI rules that apply to you in Canada
Canada has no AI law, no rule that a company must tell you when you're talking to a bot, and no regulator who can fine a company for misusing your data with AI. What does exist is smaller, mostly provincial and mostly about work. This page is the whole list, with the date each item was last checked.
Last checked: November 3, 2026
In March 2026, Statistics Canada asked workers whether they had used generative AI on the job in the past year. 35.9% said yes, and 93.4% knew what it was. So more than a third of us are using these tools at work, and most of us are guessing at the rules.
I’m not a lawyer. This is a map, not advice, and every item links to its source so you can check it. Dates are checked against LEGISinfo, Justice Canada, Statistics Canada and the Ontario Ministry of Labour guide.
Three rules people assume exist
“Canada has an AI law.”
It doesn’t. The Artificial Intelligence and Data Act was part of Bill C-27, and it died when Parliament was prorogued in January 2025. No replacement has been introduced. The bill most often described as Canada’s AI law, C-36, is a privacy bill. It would require companies to explain certain automated decisions, which matters (more on that below), but it has sat at first reading since June 15, 2026, with no second-reading debate recorded.
“A company has to tell me when I’m talking to a bot.”
No federal or provincial law requires that for adults. Bill C-34, the Safe Social Media Act, would stop regulated chatbot services from posing as human and put a duty of care on them. The House sent it to the Heritage committee on October 29, 2026. Until it passes and comes into force, the “you are chatting with an AI assistant” line you see on some sites is the company’s choice. The EU has had a disclosure rule in force since August 2026. It binds companies operating there, not here.
“The Privacy Commissioner can fine them.”
Under the current federal law, PIPEDA, the Commissioner can investigate and publish findings. The office cannot order a company to do anything and cannot impose a penalty. The one exception is a fine of up to $100,000 for knowingly failing to report a data breach. C-36 would hand enforcement to a new commission with real penalties. See item 1 for where C-36 is.
The rules that do apply to you
Organised by the situation you’re in, because that’s how the question arrives.
When you apply for a job in Ontario
Since January 1, 2026, an employer with 25 or more employees must state in any publicly advertised job posting whether it uses AI to screen, assess or select applicants. The same line has to appear on the application form. That is the whole rule. It’s a disclosure, not a limit: nothing stops the employer from using the tool, and you have no right to ask for a human reviewer.
The Ministry of Labour hasn’t said what counts as “AI” or how much detail is required, so many employers now add a blanket line to every posting. Indeed’s Hiring Lab found the share of Ontario postings mentioning AI rose from 9% in October 2025 to 28% in May 2026, as reported by Canadian HR Reporter. Internal postings and roles outside Ontario are excluded. I found no equivalent rule in any other province.
When your employer watches your screen
Also Ontario, also 25 or more employees, since 2022. The employer must have a written policy saying whether it electronically monitors employees, how, in what circumstances and for what purposes, and it must give you a copy. The policy doesn’t restrict monitoring. It doesn’t give you a right to not be monitored. The only thing you can complain to the Ministry about is not receiving it. So the useful move is simple. Ask for it, and read what it says about AI.
When a machine decides something about you
This is the one real right in the country, and it exists in one province. Quebec’s private-sector privacy law has required, since September 22, 2023, that a company making a decision about you based exclusively on automated processing of your personal information tell you so. On request it has to disclose the personal information it used and the main factors behind the decision, and let you submit observations to an employee who can review it. A loan, a rental screen, a job application scored by software: in Quebec, you can ask for the human.
Everywhere else in Canada, PIPEDA gives you a right to see and correct your personal information, and that is where it stops. C-36 would add an explanation right nationally. It’s at first reading.
When the government’s computer decides
Federal departments that use automated systems to make or recommend decisions about you, such as a benefit, a permit or an immigration file, are bound by the Treasury Board’s Directive on Automated Decision-Making. It requires an algorithmic impact assessment before the system runs, notice that automation is being used, a meaningful explanation of the decision and a way to challenge it. Systems built before June 24, 2025 had until June 24, 2026 to comply.
Two limits. It’s a directive, not a statute, so you enforce it through complaints and the courts rather than a hotline. And it doesn’t cover the Canada Revenue Agency or the provinces.
When a company’s chatbot gets it wrong
In February 2024, British Columbia’s Civil Resolution Tribunal heard the case of a passenger who followed Air Canada’s chatbot’s advice on bereavement fares and lost money. Air Canada argued the chatbot was a separate legal entity responsible for its own actions. The tribunal called that “a remarkable submission,” said the chatbot “is still just a part of Air Canada’s website,” and awarded him $650.88 in damages (Moffatt v. Air Canada, 2024 BCCRT 149).
It was a small-claims case about a company’s own customer, so it doesn’t settle the bigger questions. The principle stands, though: a company owns what its software tells you. Keep screenshots.
When someone makes a sexual deepfake of you
Since July 18, 2026, the Criminal Code offence of sharing intimate images without consent covers sexual deepfakes, and threatening to share one is a separate offence with a maximum of 10 years on indictment. That came from Bill C-16, the Protecting Victims Act, which received Royal Assent on June 18. Making a deepfake and keeping it is still not a crime, a gap law professor Robert Diab has written about. In B.C., the Intimate Images Protection Act also lets you apply to the Civil Resolution Tribunal for a takedown order and damages, and it covers altered images by name.
If this is happening to you right now, the scams and deepfakes page has the first 15 minutes and where to get help.
When you paste work into a chatbot
No AI rule, and the old rules apply. Customer or employee personal information is covered by PIPEDA, or Quebec’s law, or the provincial acts in B.C. and Alberta, whichever tool you type it into. In practice the document that governs you here is your employer’s acceptable-use policy. Read it before the tool reads your data. The AI rules at work page has the test to run before you paste.
If your employer sells into Europe
Then some of its AI is already regulated, just not by Canada. The EU AI Act’s transparency rules took effect in August 2026. Its rules for high-risk systems, which include hiring and worker-management tools, were pushed to December 2, 2027 by the omnibus regulation adopted in June. Canadian companies with EU customers are building to that standard now. The question for Canada has never been whether AI here gets rules. It’s whose.
What’s in force, what’s pending, and when it was checked
Every row was checked on November 3, 2026. If a date here is older than the one at the top of the page, nothing had moved when I last looked.
Ontario: AI disclosure in job postings
Employers with 25 or more staff. Disclosure only.
In forceSince Jan 1, 2026Ontario: electronic monitoring policy
Employers with 25 or more staff must have and hand out a written policy.
In forceSince Oct 2022Quebec: right to know and to a human review of automated decisions
Section 12.1 of the private-sector privacy law.
In forceSince Sept 22, 2023Federal: Directive on Automated Decision-Making
Federal departments only. Not the CRA, not the provinces.
In forceLegacy systems had until June 24, 2026Criminal Code: sexual deepfakes
Bill C-16. Sharing and threatening to share are offences.
In forceSince July 18, 2026Bill C-36: privacy reform, automated-decision explanations
No second-reading debate recorded.
PendingFirst reading June 15, 2026Bill C-34: social media and chatbot duties
Standing Committee on Canadian Heritage.
PendingCommittee since Oct 29, 2026Bill C-8: critical cyber systems
Operative part waits for a Cabinet order. None found as of Oct 1, 2026. Not about AI.
PendingRoyal Assent June 2026Ontario: public-sector AI regulations
The July 1, 2026 regulations cover cyber security and children’s data. AI rules not yet written.
PendingAct in force Jan 2025Federal voluntary code of conduct on generative AI
No penalties.
Voluntary46 signatories as of June 4, 2026
Sources: LEGISinfo C-36, LEGISinfo C-34, Justice Canada on C-16, Treasury Board directive, Ontario EDSTA page, ISED voluntary code.
Four things to do this week
- Ask for the monitoring policy. If you work in Ontario for a company with 25 or more people, HR has to give it to you. Read it for the word “AI.”
- Read the next job posting for the AI line. If it’s missing and the employer is over 25 people, you now know something the recruiter may not.
- In Quebec, ask for the human. If software turned you down, ask for the factors and for a person to review it. Cite section 12.1.
- Screenshot every chatbot answer you rely on. Air Canada paid because a passenger had the receipt.
This page is dated on purpose. Three items on it changed in the five months before it was written, and C-34 could move again before the end of 2026. When something moves, the date at the top changes with it.
More on the site
AI rules at work →
The EU AI Act, your company’s policy, who owns AI output, and what not to paste into a chatbot.
AI scams and deepfakes →
The first 15 minutes after a deepfake, voice-clone defences, and where to get help in Canada.
AI literacy at work →
What to learn, what to question, and what HR and L&D should teach.
AI in the real world →
Documented cases of AI misuse, and the patterns behind them.
Common questions
Does Canada have an AI law?
No. The Artificial Intelligence and Data Act was part of Bill C-27, which died when Parliament was prorogued in January 2025. No replacement has been introduced. Bill C-36, tabled June 15, 2026, is a privacy bill that would require companies to explain certain automated decisions, and as of November 2026 it has not moved past first reading.
Does my employer have to tell me if AI screens my job application?
Only in Ontario, and only as a disclosure. Since January 1, 2026, an Ontario employer with 25 or more employees must state in a publicly advertised job posting whether it uses AI to screen, assess or select applicants. It does not have to stop using the tool, and you have no right to ask for a human reviewer. No other province has an equivalent rule.
Can I ask for a human to review an automated decision about me?
In Quebec, yes. Since September 22, 2023, a company that makes a decision about you based exclusively on automated processing of your personal information must tell you, and on request must disclose the main factors and let you submit observations to an employee who can review it. For decisions by federal departments, the Treasury Board directive requires notice, an explanation and a way to challenge the decision. Elsewhere in the private sector, Canadian law gives you no such right yet.
Does a company have to tell me when I am talking to a bot?
No Canadian law requires that for adults. Bill C-34, the Safe Social Media Act, would stop regulated chatbot services from posing as human; it was sent to a House of Commons committee on October 29, 2026. Until it passes, any disclosure you see is the company's choice.
If a company's chatbot gives me wrong information, who is responsible?
The company. In Moffatt v. Air Canada (2024), British Columbia's Civil Resolution Tribunal rejected the airline's argument that its chatbot was a separate legal entity, said the chatbot was part of the company's website, and awarded the passenger damages. It was a small-claims case, but the principle stands: a company owns what its software tells you. Keep screenshots.
Is it a crime to share a sexual deepfake in Canada?
Yes. Since July 18, 2026, the Criminal Code offence of sharing intimate images without consent covers sexual deepfakes, and threatening to share one is a separate offence with a maximum of 10 years on indictment. The changes came from Bill C-16, the Protecting Victims Act. Making a deepfake and keeping it is not itself a crime.
Last checked: November 3, 2026.
The Human+AI newsletter
One email a week. Clear, skeptical, and 100% independent takes for people who work with AI. Join over 1,300 subscribers on Substack.
