A Canadian disciplinary ruling marks a turning point: it was not the use of AI that was punished, but the failure to verify — and the dishonesty that followed. Our analysis, and what we expect for the legal profession.
The disciplinary tribunal for the legal profession in Ontario has suspended a Toronto lawyer for six months and fined her CAD 10,000 (over €6,000) after she relied on case law that turned out to have been invented by an AI chatbot. According to The Toronto Star, it is believed to be the first published disciplinary decision on the improper use of AI by a lawyer. The Dutch public broadcaster NOS reported on the case on 23 July 2026.
The lawyer ran into trouble in 2025 in an estate matter. Both in her written submissions and orally in court she referred to judgments that did not exist; another authority she cited proved entirely irrelevant to the dispute, and in yet another the stated outcome was wrong. When the judge asked her to produce the sources, she could not. She ultimately admitted the documents had been drafted with the help of AI.
That was not the end of it. She initially blamed an assistant — a law student — who had supposedly done the AI work, claiming she herself did not know how such programs operate. When the regulator pressed on and she refused to name the assistant, the truth emerged: she had no assistants or interns at all at the time. Her computer's search history, obtained by the tribunal, showed that shortly after being caught she had turned to the very same chatbot again — this time asking how to prepare for her follow-up hearing and avoid disciplinary proceedings.
The tribunal does not condemn the use of AI as such. What the lawyer is reproached for is classic professional discipline: filing unverified, incorrect authorities, and then a lack of candour towards the court and the regulator. The technology is new; the standards breached — diligence and integrity — are anything but.
Nor does the case stand alone. In the Netherlands, three lawyers received a formal warning from the supervisory authority earlier this year for improper AI use in proceedings; two of them were required to complete an AI course. Worldwide, hundreds of cases have now been recorded in which courts sanctioned parties or counsel over AI-fabricated citations.
We expect supervision of AI use in the legal sector to intensify in the years ahead. What is now resolved with a warning or a mandatory course will gradually shift towards fixed frameworks: bar supervisors and disciplinary tribunals asking explicitly about AI use, courts requesting an account of the working method, and firms having to demonstrate a written AI policy — much as happened earlier with client trust accounts and anti-money-laundering rules.
At the same time, the sector itself is already building the answer. Telling is what is happening at Wolters Kluwer, the publisher behind the practice-management software Kleos that our firm also works with. Kleos now includes built-in Expert AI functionality, and since May 2026 Kleos is integrated with Libra — the legal AI workspace Wolters Kluwer acquired in late 2025. The essential difference from a general-purpose chatbot: this environment works from verified, editorially maintained legal sources, with attention to transparency and governance, so that research becomes a traceable part of the case file itself rather than a loose question to an anonymous language model.
That is precisely the direction we expect and welcome: less room for casual improvisation with invented sources, more professional environments in which the provenance of every citation can be checked. The lawyer who chooses his instruments seriously has little to fear from stricter oversight.
For legal practice in Aruba and the wider Caribbean part of the Kingdom the message is equally clear. AI can be a valuable tool — for searching case law, structuring files or preparing first drafts. But responsibility for every document filed with a court rests fully and personally with the attorney. In practice that means: every cited judgment is verified in the original source before filing, the working method is disclosed when asked, and the firm maintains clear internal rules on what AI is and is not used for. That is how we work at Glas & Glas: technology as an instrument, with human review and craftsmanship as the final word.
NOS, report of 23 July 2026 · The Toronto Star, July 2026 · Law Society Tribunal (Ontario), decision 2026 ONLSTH 136 (CanLII) · NOS, February 2026 · Wolters Kluwer, press release 19 May 2026.
This article is general information, not legal advice. Every situation is different, and we would be happy to review yours.
Tell us what you are dealing with and we will map out the legal position in clear language — with technology as a tool and craftsmanship as the standard.
Request a consultation → Send us a WhatsApp