There was a time when a sudden change in a client’s writing style raised eyebrows. A normally brief, to-the-point email that arrived sounding like a junior associate’s legal memo usually meant a spouse, sibling, or friend had stepped in to help. Today, it means something else entirely.
It means artificial intelligence.
Across Canada, AI is quietly and sometimes clumsily reshaping how people interact with the legal system. Family lawyers, immigration counsel, and appellate courts are all encountering the same phenomenon: filings, emails, and arguments drafted or heavily influenced by generative AI tools. The promise is efficiency and affordability. The reality, more often than not, is confusion, delay, and mounting costs.
The problem is not that people are using AI. That genie is long out of the bottle. The problem is how people are using it and what they expect it to be.
Many clients now treat AI as a kind of legal oracle: neutral, authoritative, and infallible. Lawyers on the ground know better. AI is persuasive, not precise. It sounds confident even when it is wrong. Worse, it has a tendency to agree with the person asking the question, reinforcing assumptions rather than challenging them. In law, where facts, context, and nuance matter deeply, that is a dangerous trait.
Courts are already seeing the fallout. Fake cases. Invented legal pathways. Citations that look legitimate but lead nowhere. In some instances, these mistakes have cost people thousands of dollars in sanctions. In others, they have put legal professionals’ reputations and even their liberty at risk.
Self-represented litigants are particularly vulnerable. Many turn to AI in good faith, hoping to level the playing field or avoid legal fees they simply cannot afford. Instead, they walk into court armed with documents that appear polished but are fundamentally flawed. Judges are then left to sift through pages of irrelevant or fabricated material, slowing proceedings and increasing costs for everyone involved.
Ironically, the attempt to save money often does the opposite. Lawyers report spending billable time reviewing AI-generated submissions that do not apply to the client’s situation at all like arguments about matrimonial property from someone who was never married. The technology did its job: it produced text. What it did not do was exercise judgment.
AI is also changing the lawyer-client relationship in subtler ways. Some clients now run their lawyer’s work through AI to “fact-check” it, sometimes uploading sensitive personal information in the process. Others second-guess professional advice because an algorithm told them something different. Trust erodes, communication breaks down, and representation becomes harder, not easier.
None of this means AI has no place in law. Used properly, it can be an excellent assistant summarizing documents, organizing timelines, and helping people understand general legal processes. Within law firms, specialized and regulated AI tools are already improving efficiency behind the scenes.
But AI is not legal judgment. It does not understand strategy, credibility, or the human dynamics that shape outcomes in courtrooms and immigration offices. It cannot weigh risks, read between the lines, or adapt advice to the messy realities of individual lives. And it certainly cannot take responsibility when things go wrong.
Some courts and professional bodies are beginning to respond, requiring disclosure when AI is used and warning of penalties for misuse. Educational efforts aimed at self-represented litigants teaching them how to verify sources and follow court rules are a step in the right direction. This is, as some advocates describe it, harm reduction.
Ultimately, the legal system must adapt, but it cannot abdicate its core principles. Access to justice is not improved by confident nonsense, no matter how well written it is. Technology should lower barriers, not create new traps.
AI is here to stay. The question is whether we treat it as what it truly is a tool or continue to mistake it for a substitute for human judgment. In the courtroom, that distinction matters more than ever.
