The Federal Court's rule turns on one test: does the AI's role in a filing resemble that of a co-author? Where it does, a Declaration is required. Where it merely suggested or critiqued, it is not.
Key takeaways
The Federal Court has addressed AI in court documents twice: a Notice to the Parties and the Profession on AI in filings, most recently updated May 7, 2024 (an earlier version dates to December 20, 2023), and a separate Interim Principles and Guidelines on the Court’s Use of Artificial Intelligence, dated September 29, 2025, governing how the Court itself uses AI. (Federal Court, Notices to the Parties and the Profession) The two documents answer different questions — one is about what litigants file, the other about what the Court does — and the language in each is worth reading directly rather than summarized.
“The Court expects parties to proceedings before the Court to inform it, and each other, if documents they submit to the Court, that have been prepared for the purposes of litigation, include content created or generated by artificial intelligence (‘AI’). This shall be done by a Declaration in the first paragraph stating that AI was used in preparing the document, either in its entirety or only for specifically identified paragraphs.” (Federal Court, Notice to the Parties and the Profession, May 7 2024) The sample the Court itself provides: “Artificial intelligence (AI) was used to generate content in this document at paragraphs 20-30,” with a French equivalent for bilingual filings.
The test is not whether AI touched the document at all — it is whether AI’s role “resembles that of a co-author.” The Notice draws the line explicitly: “a Declaration is required if content in the material was directly provided by AI… However, a Declaration is not required if AI was used to merely suggest changes, provide recommendations, or critique content already created by a human who could then consider and manually implement the changes.” (Federal Court, Notice to the Parties and the Profession, May 7 2024) Spellcheck, grammar tools and ordinary document editing sit outside the Notice entirely, by definition: it “does not apply to AI that lacks the creative ability to generate new content… including programs such as system automation, voice recognition, or document editing.”
Two categories fall outside the Declaration requirement by design: Certified Tribunal Records, and expert reports, which instead disclose AI use through the methodology summary required under subparagraph 3(i) of the Expert Witnesses Code of Conduct (Rule 52.2 of the Federal Courts Rules) — a separate, already-existing disclosure channel rather than a gap. (Federal Court, Notice to the Parties and the Profession, May 7 2024) And the duty travels with the file: successor counsel taking over a matter must “make best efforts to ascertain whether any such content has been included” in material inherited from previous counsel or a self-represented litigant, so a Declaration obligation cannot be avoided simply by changing lawyers partway through a proceeding.
The Notice defines both terms for the purposes of the Declaration regime. “Hallucination” is “facts, citations, and other content generated by AI that are not true, and have been fabricated by AI in response to a prompt or request.” “Deepfake” is “AI-generated images of human subjects that either replace one person’s likeness convincingly with that of another, or that do not exist in real life.” (Federal Court, Notice to the Parties and the Profession, May 7 2024) Having these terms defined inside an actual Canadian court document, rather than borrowed from a vendor glossary, is itself useful: it is the definition a Federal Court proceeding will actually apply.
Beyond the Declaration mechanics, the Notice sets three principles. Caution: use “only well-recognized and reliable sources… official court websites, commonly referenced commercial publishers, or trusted public services such as CanLII.” “Human in the loop”: “the Court urges verification of any AI-created content.” Neutrality: “the inclusion of a Declaration, in and of itself, will not attract an adverse inference… the party signing a document submitted to the Court bears responsibility for the accuracy and veracity of its contents.” (Federal Court, Notice to the Parties and the Profession, May 7 2024) Neutrality is the one that changes behaviour least intuitively: declaring AI use does not itself create risk, but it also does not reduce the signer’s responsibility for what the document says.
The September 2025 Interim Principles bind the Court’s own use of AI, not litigants. Its central commitment: “the Court will not use AI, and more specifically automated decision-making tools, in making its judgments and orders, without first engaging in public consultations.” Seven principles follow — Accountability, Respect of fundamental rights, Non-discrimination, Accuracy, Transparency, Cybersecurity and “Human in the loop” — with Transparency committing the Court to “authorize external audits of any AI-assisted data processing methods that it embraces.” The same page discloses a live, narrow use already underway: “AI tools are now being used to assist language specialists who translate Federal Court decisions,” described as “complementary to the translation — they do not replace the human doing the work,” under a quality-control system run by the Courts Administration Service. (Federal Court, AI in Court Proceedings)
A self-represented litigant drafts a factum with an AI tool that generated the argument in paragraphs 20 through 30 outright, while the rest was written and only lightly polished by the litigant themselves. The Declaration goes at the top: “Artificial intelligence (AI) was used to generate content in this document at paragraphs 20-30.” (Federal Court, Notice to the Parties and the Profession, May 7 2024) If the litigant later hires counsel, that lawyer inherits the successor-counsel duty to check whether any further AI-generated content is buried in the file. And if paragraphs 20 through 30 turn out to cite a case that does not exist, the Neutrality principle means the Declaration itself provides no shelter — the signer still bears responsibility for the citation being real.
The Declaration regime is additional to, not a replacement for, the ordinary rules on what makes a document reliable evidence in a Canadian proceeding — rules that already existed before generative AI and still apply to how a document is authenticated and disclosed generally. (treadstonelaw.ca, evidence in Ontario civil proceedings) And the Federal Court is not writing in isolation: the Law Society of Alberta’s survey of Canadian AI guidance counts the Federal Court alongside eleven courts and nine law societies that have each issued their own document, none of them a ban. (LSA, Gen AI Rules of Engagement)
Related: Alberta courts and AI-generated material, what happens when AI invents a case citation, and can lawyers use generative AI in Canada
No. The Notice explicitly does not apply to AI that lacks the creative ability to generate new content, and names system automation, voice recognition and document editing as examples outside its scope.
No. The Court's Neutrality principle states that including a Declaration will not itself attract an adverse inference, but the party signing the document still bears responsibility for the accuracy and veracity of its contents.
Not without first engaging in public consultations, per its September 2025 Interim Principles — and even then, the seven principles bind that use to accountability, non-discrimination, accuracy, transparency, cybersecurity and human oversight.
The same co-author question shows up in due diligence on any document set an AI tool helped assemble.
