Since 1 January 2026, an Ontario employer with 25 or more employees has to say so in a publicly advertised job posting when artificial intelligence is used to screen, assess or select applicants. Here is what that covers, and what it does not.
Key takeaways
A publicly advertised job posting is defined broadly: an external posting that an employer, or someone acting for the employer, advertises to the general public in any manner. If artificial intelligence is used to screen, assess or select applicants for that posting, the posting has to include a statement disclosing it.
Ontario defines artificial intelligence for this purpose as a machine-based system that, for explicit or implicit objectives, infers from the input it receives in order to generate outputs such as predictions, content, recommendations or decisions. That is a wide definition, and it is wide on purpose. It is not limited to something marketed as AI.
The job-posting requirements do not apply to an employer with fewer than 25 employees on the day the posting is published. The count is of individual employees, not full-time equivalents, so a part-time or casual employee counts as one person regardless of hours.
The threshold is measured at the moment of posting, which matters for seasonal businesses that cross 25 people at some points in the year and not others. If you are close to the line, the safer practice is to keep the disclosure in your posting template year-round rather than recount every time a role opens.
Most employers do not deliberately buy an AI screening product. They inherit one. Resume parsing, candidate ranking, knock-out question scoring and automated assessment marking are built into mainstream hiring tools, and any of those steps can influence which applicants a human ever sees.
The practical exercise is to walk one req end to end and write down every point where software reads, scores, ranks or filters an application, whether that happens in your applicant tracking system, on a job board, or inside an assessment vendor. Where a step is genuinely mechanical, such as de-duplicating two copies of the same application, it is unlikely to be screening. Where a step changes who gets looked at, treat it as in scope and disclose.
The AI disclosure did not arrive alone. Postings in scope also have to include expected compensation or a range, and Ontario limits how wide that range can be, with an exemption for roles paying above a set threshold. Requirements for Canadian experience are prohibited in postings and in associated application forms.
Two further obligations affect the back end of the process rather than the advertisement. Employers have to inform applicants they interviewed of a hiring decision within 45 days of the last interview, and have to keep copies of postings and application forms for three years after public access to the posting is removed. Both are exactly the kind of deadline-and-record work that an automated process handles better than a busy person does.
Add the disclosure sentence to your posting template so it is present by default and removed deliberately, rather than the other way round. Ask your agency, in writing, whether it uses AI to screen candidates it submits to you, because their screening sits inside your obligation.
Then make the record-keeping automatic. If postings, application forms, interview dates and outcome notices are all captured as a by-product of running the process, the three-year retention and the 45-day notice stop being things anyone has to remember.
A 30-minute call is enough to tell you whether AI pays for itself here.
