A Canadian project is launching an open, transparent AI-detection tool to help courts assess AI-generated content, addressing flaws in current commercial tools that are opaque, unreliable, biased against non-native English speakers, and prone to false positives.

The initiative emphasizes transparency, accuracy, and the provision of confidence levels and explanations for its decisions, moving beyond simple binary flags.

The effort targets gaps in court readiness for AI evidence and seeks to be accessible to courts and self-represented litigants across North America, not just Canada.

Lead technical work will focus on keeping pace with rapidly evolving AI content, requiring ongoing updates and adaptation.

The Canadian Press published the report on December 5, 2025.

Funding comes from the Canadian AI Safety Institute, with two inaugural initiatives receiving $700,000 over two years, aligning with Canada's AI safety strategy.

The project brings technologists and legal scholars together from Ontario and British Columbia to ensure accessibility for courts and self-represented litigants across North America.

An advisory board comprising judges and self-represented litigant support groups guides the project for broad North American adoption.

A minimum viable product is anticipated in about two years, with the understanding that ongoing funding and iteration will be necessary beyond the grant period.

The team plans to deliver an open-source, free tool designed to help courts detect AI-generated content and distinguish real from manipulated evidence, including images and videos.

The project responds to the real-world need for affordable, rapid, and transparent AI-content verification to prevent miscarriages of justice from manipulated media.

Crucially, the tool must quantify confidence, provide explanations, and avoid a simplistic verdict to gain judicial trust.