Rainy day reflections on AI and surreptitious recordings
A brief discussion of unreliable AI and a case comment on admissibility of surreptitious recordings
Here in the National Capital Region we finally received the rain we needed - although it made for a rather gloomy Thursday.
I have been immersing myself in learning about Artificial Intelligence (AI) as I prepare for two presentations in November (more details in the next newsletter). I refuse to use AI for writing but I have dabbled a bit in using it for research - mostly for my other vocation - as a jazz DJ at CKCU FM. This week I was doing research for my show that aired on Wednesday on tenor sax player Billy Harper. I was using Co-Pilot, the AI that comes with the browser Microsoft Edge. I realize that this is not the AI tool that lawyers and professionals would routinely use - but it is one that might easily be used by self-represented parties. This is troubling. I was asking Co-Pilot for easily verifiable information - in this case, albums that featured solos by Billy Harper. Co-Pilot directed me to albums that he did not appear on - and even described his “soaring” playing on these albums. Of course, I received a profound apology each time I corrected it. I do know how to research discographies and verify information - but someone who does not know how to go about verifying information would likely rely on this erroneous information.
The embarrassment of referring to an album with the wrong lineup is not, in the scheme of things, a big one (although it could affect one’s reputation as a DJ). The danger is using AI for more important or consequential decisions.
The Saskatchewan Access to Legal Information project has prepared a guide to using AI for legal research. It emphasizes the importance of the prompts used:
How a question is worded can impact the accuracy and quality of AI-generated information. Well-structured prompts (questions) may reduce, though not eliminate, the risks.
Melanie Hodges Neufeld recently documented her efforts to obtain legal information about the Saskatchewan Assessment Process for challenging a lawyer’s account using ChatGPT-5. I recommend looking at the prompts she used and the subsequent refining of those prompts. Her conclusion was that, although an improvement over earlier versions of ChatGPT, gaps were still evident. Her conclusion was:
Based on my experiment and those of others …, the content generated by ChatGPT cannot be the only step involved in developing a legal information resource for the public. … It is a good starting point, not the end product.
ChatGPT does not agree - today I got a marketing email that reads, in full:
You don’t need a plan or the perfect question to use ChatGPT.
Just start typing whatever’s on your mind — a half-baked idea, a random question, a weird “what if.”
ChatGPT is built for all of it.
No pressure. No perfect prompts. Explore ideas, create images, write stories, and summarize files in just a simple conversation.
Admissibility of surreptitious recordings
In a recent Federal Court of Appeal decision, the court addressed the decision of the Canada Industrial Relations Board not to admit as evidence surreptitious audio recordings that an applicant made (Fearing v. Canada Council of Teamsters, 2025 FCA 167). The court noted that the Board can reject evidence “in its discretion” and “as it sees fit” whether or not admissible in a court: Canada Labour Code, R.S.C. 1985, c. L-2, s. 16(c).
The court stated that the Board drew on its labour relations experience and expertise and its earlier decision in Valenti and CUPW, Re, 2018 CIRB 866 at paras. 8-11, in concluding that the regular admission of surreptitious recordings would be damaging to labour relations. The court found the ruling to be reasonable. However, the Board has not posted its original decision online - so we cannot see their reasoning.
The Valenti decision gives us a sense of how the Board approached the request in this case. In Valenti, the Board referred to an earlier Board decision for the rationale for usually excluding surreptitious recordings (D.H.L. International Express Ltd. (1995), 99 di 126):
It must be remembered that parties who appear before the Board typically continue in an ongoing labour relations relationship with one another. The successful functioning of that relationship is dependent, as far as possible, on mutual trust and respect. It is difficult to imagine how open and frank discussions, in an atmosphere of mutual trust and respect, could be carried on if either party was concerned that the other might be recording the conversation to be played back to the Board or in another forum at some subsequent period of time.
Were the Board to adopt a broadly permissive policy with respect to the admission of surreptitiously recorded evidence, it is not difficult to envisage how proceedings before it could become inexorably protracted by applications to have the recordings in question properly proved in an evidentiary fashion similar to that in the courts. Nor is it difficult to anticipate adjournments requested in order to permit the tapes to be analyzed by experts, not to mention the introduction of expert testimony relating to the recordings in question. Without attempting to overstate the case, to allow this type of evidence without restriction, would open an evidentiary Pandora’s box from a labour relations perspective. This is particularly so when one keeps in mind the objects and purpose of the Code and the Board’s role in the implementation of the same.
(pages 137–138; and 310)
In Valenti, the Board noted the ubiquity of mobile devices and the challenges that would ensue if such evidence were to be “submitted freely and without restriction”. It stated that admission “would create an atmosphere of mistrust, which would not be conducive to having open discussion or resolving labour disputes”. The Board noted that each case must be examined on its own circumstances. It set out a non-exhaustive list of criteria to consider in determining whether to accept such evidence:
• The burden of proof that must be met;
• The adverse effects of the recordings on the parties’ labour relations;
• The reliability of the audio evidence;
• The parties’ ability to present testimonies and thus allow for cross-examinations;
• The need to ensure a fair process; and
• The need to ensure full disclosure of the evidence, which will in turn promote the timely resolution of the matter.
The Board also noted that the party seeking to introduce such evidence would be required to demonstrate “that the same evidence cannot be obtained through other means and that it is of such probative value as to outweigh any negative or prejudicial effect it will have on the process or on the relationship between the parties”.
In 2006, I was faced with a request for admission of a surreptitious recording of a supervisor. I found that the evidence was not relevant to the issue before me (jurisdiction) but I did note that, “[g]enerally, the surreptitious recording of conversations in the workplace should not be encouraged”. In a 2007 decision, I also ruled against the admission of surreptitious recordings of settlement discussions. I held that admitting the recordings would be contrary to labour relations policy, “as the surreptitious recording of settlement discussions should not be encouraged”. In addition, the participants in the settlement discussions were available to be called as witnesses.
