Law360 Canada (September 21, 2026, 5:15 PM EDT) -- Judges in Atlantic Canada are confronting a rapid rise in the use of artificial intelligence in court proceedings, warning that unverified AI-generated materials, deepfakes and increasing volumes of submissions could threaten accuracy, privacy and the efficient administration of justice.
On Sept. 17, Justice Peter Browne of the Supreme Court of Newfoundland and Labrador and Justice Maya Hamou of the Court of King’s Bench of New Brunswick participated in a webinar discussing emerging issues of AI use in Atlantic courts.
“What strikes me … is how quickly the discussion about AI has become real,” said Justice Browne, in noting changes about the AI conversation in the last 12 to 18 months.
He said that a few years ago, the notion of AI was largely theoretical, and now AI-assisted materials are coming up in court files, especially when it comes to self-represented litigants (self-reps).
There are real risks pertaining to reliability and privacy, he noted, but there are also emerging opportunities in court administration, including accessibility for litigants and information management. The question is no longer how AI will affect the courts — it’s here, and the question is how the courts will deal with it.
Justice Hamou added that in the last few weeks and months, the landscape has changed again. Tech CEOs are speaking up about the need for guardrails. The discussion is evolving rapidly, and courts need to respond to that.
“This profession perhaps has been a little reluctant to embark on the AI discussion.… But it’s also because of the nature of what we do as lawyers, as judges, as law students, as law professors,” she said.
“Our role, in part, is to uphold the rule of law, the administration of justice. We have a duty to make sure it continues to function properly and isn’t taken over by these outside sources.”
Professionals in the field could discuss how to use AI in a controlled way for tasks, with restricted access and oversight of its output, she suggested. Systems can be created to assist while maintaining the rule of law and protecting the administration of justice.
Justice Hamou said there has been a sharp increase in the use of AI in the courtroom, and not all of it is disclosed. Self-reps are commonly using AI to prepare submissions, even orally.
With lawyers, it’s less obvious, but one giveaway is them becoming more eloquent the next day with a sharp turn in the way they express themselves. They might be unable to engage with the materials they submitted or unable to discuss the case they referred to.
With AI use, the court receives voluminous submissions and lengthy issues, leading to an “explosion of paperwork” in the courts, she said.
Judges will address AI use with the parties if they notice it, she said. The New Brunswick Court of King’s Bench is also developing policies on AI use by participants in the justice system. She said people need to look out for potential dangers, including being fooled by polished language.
“Ultimately, the AI is not creating a product that a lawyer may have produced, or an articling student may have produced. It won’t be of the same quality because it lacks that human component that is integral to our justice system.”
Justice Browne noted that there is no positive obligation to disclose the use of AI, but its use tends to be obvious. When lawyers do it, he assumes that they’ve verified the content and adjusted it accordingly.
Accuracy must always be the primary principle, he said, and self-reps oftentimes are not able to understand the nuances of the issues in their claim when using AI.
Justice Hamou stated that problematic use of AI is when pleadings are not reviewed or adjusted. In some circumstances, preparing a chronology with AI may be a positive use.
Another problem is people citing hallucinations to discredit AI systems. She said this takes away the legal professional’s role to review the information and assess credibility, reliability and where the information comes from. The focus should instead be on competency.
“That citation was improper, but did you check it? Because that’s the fundamental question here,” she said.
“It’s not that the materials were presented as false or misleading. It happens in other contexts … [but] the onus is on the parties when they present materials to the court. They have a responsibility to make sure it’s accurate.”
She added that AI can’t do legal analysis the way humans can, and the human aspect of the decision-making process is important to the administration of justice.
Justice Browne said that “one of the big, scary doors for us to open” is deepfakes and how to recognize them.
“As a trial judge, assessing evidence is critical, and credibility is critical. The pace in which deepfakes are becoming more and more realistic is concerning.”
While he has not come across them himself, his family law colleagues have raised concerns emerging in family law proceedings in which a spouse might try to use a deepfake to attack the credibility of another spouse. Such an aspect will also potentially add more trial time.
“As a judge, I’m not an expert, so if there’s a challenge to whether a particular exhibit is a deepfake or not, I’m going to need to rely on some expert opinion,” he said.
“So now we add another layer to that, with requirement for an expert evaluation as to whether or not the exhibit or the document is a deepfake, and that only adds to issues like backlog and delay.”
Another thing he noted was that the court is now experiencing litigants wanting to wear Meta glasses to record what is happening in the proceeding. One litigant told him the glasses were not turned on, and he took it at face value.
Justice Browne noted that a policy statement has been issued from Newfoundland and Labrador’s law society to ensure everything submitted to the court is verified, despite there being no positive obligation to disclose AI use.
He said that courts are not digitized, and discussions on AI can’t occur unless that platform is there. AI policies for the court can’t take hold unless that threshold is reached.
Audiovisual platforms in the province’s courts are expected to be implemented in the next six months. Down the road, there may be some online scheduling, unofficial AI transcripts and translations.
In New Brunswick, there are varying viewpoints among the bench, with some people believing there should be no AI integration in the court at all, noted Justice Hamou. This may come down to what people are submitting before the court.
One concern is that legal tests could be changed... A change in a word to simplify a sentence could change a legal test, she said. If the judge doesn’t have the instinct to verify the source, suddenly a new test is appearing in the jurisprudence.
She added that the Canadian Judicial Council plans to come out with new guidelines regarding AI before the end of the year. The last time orders were issued on AI was in October 2024, for judges to not delegate any part of the decision-making process to AI.
Once the new order comes out, the Court of King’s Bench will be in a position to issue its guidance on AI use. She noted that industry should be careful not to mix AI use by members of the judiciary and how people who come before the courts can use AI.
Justice Browne said they may observe what happens in other jurisdictions and learn before adopting and creating any policies and following the council’s upcoming guidelines.
In the near term, he said, there are potential administrative scheduling accessibility benefits that can be harnessed, and after digitization, the technology can merge and provide valuable assistance to both litigants and judges without affecting decision-making.
Justice Hamou noted the pilot project in the Quebec Superior Court on its first documented and controlled experience using AI integration for tasks within specialized parameters.
Eighty-four per cent of judges affirmed that they could use these agents without undermining their professional autonomy or judgment, and that the technology was compatible with judicial functions.
“Talking about AI generally in the ChatGPT, Claude sense, is scary and also dangerous,” said Justice Hamou. “So controlled environments, specified tasks — that’s where the discussion has to take place.”
Justice Browne noted cybersecurity as another concern for potential privacy and confidentiality issues, as the cloud-based storage systems would need to be located in Canada, so that sensitive information is protected when using AI.
“I would hopefully never see or want to see a situation where AI takes over the decision-making process. It must remain a human decision,” he said.
“If AI were to take over decision-making, would it be able to colour the law as it will evolve moving forward? It’s a bit of a philosophical question,” added Justice Hamou.
“But I like to believe that the judges of these courts across Canada … have evolving views as the country evolves. So, if we lose sight of what makes the law what it is, I don’t think it’s going to work,” she said. “I hope not to see AI take over because then it’s just a rule book, and that’s not how the law works.”
AI often has black-and-white answers, she said, but when peeling back layers in the law, it tends to be grey, which is the human loop aspect. It’s possible that a professional would disagree with an AI-given interpretation on the cases it puts forward. As technology evolves, the field will need to constantly adjust.
On AI education in the legal field, she noted that the Canadian Bar Association has a program to learn about AI and develop competency.
“I think AI is going to have a place in our justice system,” she said. “We just need to carefully surround where that’s going to go and what the parameters around it are going to be.”
“Educate yourself on these AI tools, and remember that ultimately you are the lawyer,” she added. “You are expected to have the confidence to verify the information you provide to us.”
Disclaimer: This webinar was hosted by LexisNexis Canada, the publisher of Law360 Canada.
If you have information, story ideas or news tips for Law360 Canada on business-related law and litigation, including class actions, please contact Anosha Khan at anosha.khan@lexisnexis.ca or 905-415-5838.