Being let go for using an AI tool is judged the same way any other policy breach is — there is no separate, harsher standard because the tool involved happens to be an AI.
Short answer
Yes, it's possible — but it runs through ordinary dismissal law, not a special AI rule. Any employer can end an employment relationship without cause by giving notice or pay in lieu. Ending it without notice, on the basis of the ChatGPT use itself, requires the conduct to clear Ontario's narrow test for “wilful misconduct, disobedience, or wilful neglect of duty that is not trivial and has not been condoned by the employer.”
Ontario's ESA guarantees minimum notice or termination pay to almost every employee, and it withholds that guarantee for only a short list of exceptions — chief among them employees “who are guilty of wilful misconduct, disobedience, or wilful neglect of duty that is not trivial and has not been condoned by the employer.” A single instance of pasting a routine question into a public chatbot, with no confidentiality breach and no prior warning, is a long way from that bar. A deliberate, repeated breach after being told to stop — especially one that exposes client or personal information — is a different case entirely.
For a policy to justify discipline, it generally has to have been a real, communicated part of the workplace's rules before the conduct happened. Ontario courts have treated broad, one-sided language letting an employer “change any term ‘at its sole discretion’ without the employee's agreement” with skepticism, particularly for anything reaching a fundamental term of employment. An ordinary conduct policy — unlike a cut to pay or duties — usually falls within an employer's normal ability to set workplace rules, but it still has to have been put in place and actually communicated. A written policy that is actually communicated matters a great deal to whether breaching it later can fairly be called misconduct at all.
The AI tool itself isn't what changes the analysis — what a person did with it does. Using an unauthorized AI tool to draft an internal memo is a minor policy slip. Pasting a client's confidential file into a public AI tool, after being told not to, is a different order of problem — not because AI is involved, but because it's a serious, repeated, informed breach of a real rule, which is exactly the shape of conduct the wilful-misconduct test is built to catch.
See also: where a rule like this should actually live before it's ever enforced, what happens when staff adopt AI tools nobody approved.
Setting the rules staff actually work under once a tool is live day to day is what ai-operations covers.
