Blog

New Civil Procedure Rules for AI-Generated Evidence Arriving Soon

the intersection of law and tech - glowing scales of justice

The artificial intelligence (AI) genie may well and truly be out of the bottle. But a new framework developed with human intelligence should ensure that, with appropriate safeguards, our profession can preserve and protect trust in our courts as they increasingly encounter AI-generated evidence.   

The Ontario Civil Rules Committee (CRC), chaired by Court of Appeal Justice Peter Lauwers, recently submitted a proposal for revised rules on AI-generated evidence to Ontario’s Attorney General Doug Downey for review and approval. With finalized text expected imminently, it’s worth exploring what our profession thinks about this topic and why the CRC came to its decision.

In this blog post, I outline work the CRC subcommittee on artificial intelligence undertook in advance of the pending rules changes. As a member of this subcommittee, I agree with its recommendations and the path forward that we are proposing: regulation over restriction.

Protecting Integrity in Adjudication and the Development of Common Law.

The rapid advance and deployment of AI in the past few years has been staggering. While the earliest large language models left much to be desired, successive updates by the major AI developers have vastly improved what AI can accomplish when a user is adept at employing prompts.

The results are as impressive and they are unnerving. AI-generated text has achieved graduate levels of sophistication, and AI-generated images and sound recordings are becoming increasingly difficult to discern from real reproductions.

As a result, in addition to the early problems brought about by AI-generated evidence (hallucinated citations, training bias), other concerns (deep fakes, AI agent coordination and autonomy) are quickly coming into view on the horizon.

But proposing to restrict a technology that has already become deeply integrated into our lives was not something our subcommittee believed to be viable or even preferable. Instead, we drafted a series of proposals that seek to regulate AI-generated evidence in a manner consistent with the long-standing principles and values of our judicial system.

Consultation on Proposals.

In 2025, the CRC AI subcommittee presented four proposals for consultation among civil litigators:

  • A definition of AI
  • Requirements for parties who seek to put forward evidence generated by a computer system using AI 
  • A process for a party to challenge the authenticity of evidence generated or modified by a computer system using AI, and a determination that such evidence be deemed inadmissible unless “the proponent demonstrates on the balance of probabilities that the evidence’s probative value exceeds its prejudicial effect”.
  • A process for the admissibility of outputs of a computer system using AI in an expert report. 

Although the subcommittee noted that aspects of its proposals supplemented the Ontario Evidence Act and/or the Canada Evidence Act, consultations suggested some people in the profession disagreed. For example, the Toronto Lawyers’ Association (TLA) contended that “Ontario’s current civil rules – along with case law and legislation – can already be interpreted so they apply to cases involving AI-generated evidence”. The TLA, while heartened that this consultation was taking place, sought more insight from the committee into how and why these proposals were needed to deal specifically with AI.

In contrast, other lawyers in the profession whose casework regularly involves AI-generated deepfakes have expressed concerns over the inadequacy of current evidentiary laws in this regard. For example, in an article for Canadian Lawyer, Kelly Friedman, a litigator at Heuristica who specializes in e-discovery, privacy, and data breach issues, stated that Canada’s laws are ill-equipped to address deepfake-related risks. “The evidence law as it’s written right now doesn't quite fit the current level of technological expertise out there that we can really fake anything”, she told the publication.

The Pillars of Judicial Control.

In an article for the Law Times, Justice Lauwers, who serves as both CRC chair and AI subcommittee chair, explained why the complexity of AI requires more encompassing rules, and reaffirmed the role of judicial discretion in assessing how the rules should apply in particular situations.

“The premise is that judicial control, rather than prohibition, is the practical response to AI”, he said of the proposals. “Courts should demand enough transparency to test AI-generated material while preserving privilege and preventing fabricated evidence from distorting the adjudicative process.”

Therefore, when there is proactive disclosure of AI-generated evidence, courts may require specific information to assess whether the output could be subject to bias (particularly sampling bias and discriminatory bias prohibited by the Ontario Human Rights Code). 

Counsel may be required to identify the AI system employed, how the system was trained, the inputs and prompts they used, and evidence to support or corroborate its validity and reliability. However, judicial discretion is a key component for this proposed rule. For example, certain prompts could potentially reveal litigation strategy (which may be protected under solicitor-client privilege). Justice Lauwers notes that judges will not force disclosure of prompts, but may not admit the evidence without them if they are concerned that such evidence could “be distorted to the point of being unreliable or invalid”.

In contrast, when AI-generated evidence is undisclosed, the subcommittee proposed a rule of presumed inadmissibility if a challenge or late admission reveals the use of AI. In effect, undisclosed AI-generated evidence would be deemed inadmissible unless the evidence’s probative value exceeds its prejudicial effect on the balance of probabilities.

What’s Next?

The specific wording of the rules is embargoed until the Minister approves the finalized language. However, Justice Lauwers has advised both Superior Court and Ontario Court judges to use the draft rules as a practical guide in the meantime.

Even when the new rules come into force, the profession must remain ever vigilant of unforeseen developments in this emerging technology or in how the rules are applied in future case law to assess whether they will be sufficient or require further refinement or reevaluation.

As a proponent of mindfully adopting emerging technologies where there are clear benefits, I hope my colleagues will join me in seeing these new rules as ensuring that our justice system has the proper controls in place to employ and interact with this technology as intended, rather than dismissing its place in litigation and in our courtrooms completely.

Share

Subscribe to our Newsletter

Sign me up