Law360 Canada (October 26, 2026, 1:13 PM EDT) --
 |
| Connie L. Braun |
The invoice looks convincing. The supplier’s logo is familiar, the amount seems reasonable, and the email politely explains that banking details have changed. It even urges payment before Friday. In this fictional Canadian workplace, an AI system notices what the accounts payable team has missed: the new bank account also appears on invoices from two supposedly unrelated suppliers. A staff member calls the supplier using an established telephone number. The supplier knows nothing about the change.
This is one of AI’s most useful roles in fraud detection: finding connections that give people a reason to investigate. AI does not replace judgment; it helps direct attention. Its value increases when organizations know how to turn an alert into reliable evidence and understand what legal consequences may follow.
Finding patterns that people might miss
Fraud often hides among legitimate transactions. One unusual payment may attract little attention; a pattern of unusual payments tells a different story. Machine-learning systems can compare activity with established patterns and flag transactions for review. An organization might investigate repeated payments just below an approval threshold, sudden changes in account activity or multiple customers sharing suspicious identifying information.
tovovan: ISTOCKPHOTO.COM
Canadian financial regulators recognize this application of AI. A
report by the Office of the Superintendent of Financial Institutions and the Financial Consumer Agency of Canada identifies fraud detection, network anomaly detection and identifying suspected phishing messages among AI’s most beneficial uses in financial institutions. These systems help prioritize attention so that an unusual transaction is identified and investigated. The Canada Revenue Agency also describes using machine learning and AI to detect non-compliance and suspicious activity, supporting prevention, verification and, where appropriate, criminal investigations. That does not mean an algorithm decides that a taxpayer has committed a crime. It means technology can help identify matters requiring closer examination.
Looking more closely at documents and identities
Forgery detection presents a related challenge: determining whether a document is authentic and whether the person presenting it is authorized to use it. AI-enabled identity systems can compare photographs and assess identity documents. Royal Bank of Canada reports using AI-enabled government identity-verification technology to strengthen fraud protection during account opening. FINTRAC guidance likewise describes tools that assess document authenticity and compare a person’s selfie with the photograph on an authenticated government-issued identification document. These are aids within a verification process, not substitutes for one.
The same investigative principle applies to suspicious invoices, receipts and contracts: compare the document with independent records. Does the issuer confirm it? Does the transaction appear in the accounting system? Does an original electronic file exist, with a credible history of its creation? A polished document can contain a lie. An untidy scan can be completely legitimate.
Alerts are the beginning of an investigation
Imagine that software flags a loan agreement because a signature appears inconsistent with earlier documents. Several explanations are possible. The person may have signed on a different device, an authorized signing process may have changed, or someone may have forged the signature. The alert cannot resolve those possibilities by itself.
A reliable response would preserve the original file, obtain the signing records, contact the purported signatory through a verified channel and seek forensic assistance where appropriate. Investigators should also record what the system flagged, which version was used and how a human reviewer assessed the result. Those steps matter if the dispute reaches court. Under s. 31.1 of the
Canada Evidence Act, a party seeking to introduce an electronic document must provide evidence capable of supporting a finding that it is what it purports to be. Authentication is one evidentiary requirement, but it does not establish that every statement in the document is true. Saying only that “the software suggested it was suspicious” leaves substantial work unfinished.
What legal outcomes can follow?
Canadian criminal law already addresses fraud and forgery, regardless of whether technology helps uncover them. Under s. 380 of the
Criminal Code, fraud involving more than $5,000 carries a maximum sentence of 14 years’ imprisonment. That is a statutory maximum, rather than a prediction of the sentence in any particular case.
Forgery under s. 366 involves knowingly making a false document with the required intent that it be treated as genuine or induce someone to act. Materially altering a genuine document can qualify. Section 367 provides a maximum of 10 years’ imprisonment when prosecuted by indictment. Separately, s. 368 addresses conduct such as knowingly or believingly using a forged document as genuine, also with a maximum of 10 years on indictment.
A Quebec case involving fabricated charitable donation receipts illustrates the point. The accused pleaded guilty to forgery and fraud over $5,000 and received a conditional sentence of two years less a day. This example is a reminder that the technology used to uncover misconduct does not determine the legal outcome; the underlying conduct, the evidence and the applicable sentencing principles do.
Financial recovery is another possible outcome. A criminal court may order restitution for qualifying financial losses, but an order to pay does not guarantee that the money will actually be recovered. Organizations should therefore treat prevention, rapid response and evidence preservation as complementary, not interchangeable, steps.
Civil proceedings offer a third route. In
Hryniak v. Mauldin, 2014 SCC 7, the Supreme Court described civil fraud as requiring proof of a false representation, knowledge or recklessness about its falsity, reliance that induced action and resulting loss, on a balance of probabilities. AI may help organize evidence relevant to those elements, but the claim still requires proof.
Fraud prevention must respect privacy
Fraud detection often involves sensitive financial records, identity documents and biometric information. Collecting more information creates responsibilities of its own. The Office of the Privacy Commissioner of Canada advises organizations using biometrics to assess necessity, effectiveness, less intrusive alternatives and proportionality. Sensitive biometric information generally calls for express consent, subject to applicable legal exceptions. The guidance also stresses accuracy and error rates. Organizations should test their systems, investigate false alerts and provide meaningful human review before serious accusations or adverse decisions follow.
Conclusion
Used thoughtfully, AI can make suspicious activity easier to find. Careful investigation is what turns a signal into useful evidence, and reliable evidence is what makes accountability possible. Sometimes the most valuable outcome is the least dramatic one: a verified telephone call, a cancelled transfer and an ordinary Friday.
Connie L. Braun is a product adoption and learning consultant with LexisNexis Canada.
The opinions expressed are those of the author(s) and do not necessarily reflect the views of the author’s firm, its clients, Law360 Canada, LexisNexis Canada or any of its or their respective affiliates. This article is for general information purposes and is neither intended to be nor should be taken as legal advice.
Interested in writing for us? To learn more about how you can add your voice to Law360 Canada, contact Analysis Editor Peter Carter at peter.carter@lexisnexis.ca or call 647-776-6740.