Building tool to help courts spot AI-generated evidence is ‘challenging’: researcher

A researcher based at the University of Waterloo and York University's Osgoode Hall law school is working on a tool to help courts identify evidence created or altered by artificial intelligence. Maura Grossman, the co-director of the team, says the project is challenging, despite initial optimism about the feasibility of an effective tool.

The team developed an open-source, free tool that utilizes state-of-the-art algorithms working in unison to determine if content is AI-generated. However, real-life testing revealed difficulties in achieving consensus among the algorithms, making it a "wicked technical problem" to overcome.

Grossman, who teaches both at the University of Waterloo and York University's Osgoode Hall law school, emphasizes the complexity of the task. Courts, she believes, aren't presently prepared to handle AI-generated evidence, as existing commercial tools are not transparent or reliable enough for judicial contexts. She stresses that the tool should not be seen as a silver bullet, but rather as one piece of a broader solution that includes policy considerations and educating people about how to distinguish real from fake content.

The project is one of two initiatives funded by the Canadian Institute for Advanced Research's Canadian AI Safety Institute program, launched by the federal government in 2024.