In 2026, artificial intelligence is no longer an emerging technology at the margins of legal practice. It is embedded in how lawyers research, draft, summarize, analyze, and communicate. An estimated 10 per cent of humanity now uses AI tools weekly, while ChatGPT alone reportedly receives approximately 4.5 billion monthly visits and 2.5 billion prompts per day. The reality is simple: AI is already in our practices - whether formally approved or not.
The rise of “shadow AI” presents a particular challenge for public lawyers and regulated professionals. Shadow AI refers to the use of AI tools without organizational approval, oversight, or governance. Recent research conducted by BlackFog found that 49 per cent of workers report using AI tools without employer approval, often through free public platforms that may expose sensitive data. Gartner reported in late 2025 that 59 per cent of organizations suspect or have evidence that employees are using prohibited public generative AI tools. Alarmingly, only 47 per cent of organizations have implemented dedicated security controls for generative AI. The pattern is clear: employee AI adoption is significantly outpacing organizational governance and risk controls.
The benefits of generative AI are significant. AI can accelerate legal research, summarize lengthy records, draft first versions of memoranda and correspondence, improve plain-language communication, and automate repetitive administrative work. In resource-constrained environments, these efficiencies are attractive and, in some cases, necessary.
However, AI also carries material professional, ethical, and governance risks. Hallucinations, fabricated but convincing information, remain a persistent problem. One independent database tracking AI hallucination cases reported more than 1,300 legal matters involving hallucinated content as of May 2026, including 144 Canadian cases. Courts in Canada have already sanctioned irresponsible reliance on AI-generated authorities and submissions.
Confidentiality risks are equally serious. Public AI systems may retain user inputs, creating potential exposure of privileged, personal, or institutionally sensitive information. Deepfakes and synthetic evidence further complicate evidentiary reliability and authenticity. Even genuine evidence may now be challenged as potentially fabricated — the so-called “liar’s dividend.”
Importantly, AI does not lower professional standards. It heightens them.
The Law Society of Ontario has made clear that existing professional obligations fully apply to AI-assisted work. Competence under Rule 3.1-2 requires lawyers to understand the technologies they use and the risks associated with them. Confidentiality obligations under Rule 3.3 extend to information entered into generative AI platforms. Supervision obligations also remain unchanged: lawyers cannot delegate professional judgment to technology. For more information, review the LSO White Paper on Licensee use of generative artificial intelligence.
The emerging principle is straightforward: competent legal professionals must be AI literate before using AI in practice. AI literacy means more than knowing how to write a prompt. It requires understanding the technology’s limitations, verifying outputs independently, assessing confidentiality and bias risks, ensuring appropriate supervision, and exercising professional judgment at every stage.
AI can strengthen legal practice. Used carelessly, it can undermine public trust, professional independence, and the integrity of the legal profession. The question is no longer whether lawyers will use AI. The question is whether they will do so competently, ethically, and responsibly.
About the Author
Soussanna Karas is the General Counsel at Wilfred Laurier University (WLU), where she leads the Office of Legal Services and Fair Practices, Privacy Office and the Office of Human Rights and Conflict Management. Soussanna’s legal practice spans over 20 years, with a strong track record of providing strategic advice, leading teams, and managing complex legal and regulatory matters across multiple sectors.
Soussanna is an active voice on AI governance, professional ethics, and technology-enabled regulation. She regularly speaks to regulators, legal professionals, and industry experts on the risks and opportunities of AI. Soussanna teaches Information and Privacy Law at Ontario Tech University and contributes to OsgoodePD programs.
Before joining WLU, Soussanna served as in-house counsel, prosecutor, and litigator for several regulatory bodies. She holds both LL.B. and LL.M. from Osgoode Hall Law School.
Any article or other information or content expressed or made available in this Section is that of the respective author(s) and not of the OBA.