CL Talk

Beyond hallucinations: the biggest mistake law firms can make in implementing AI

Jan 05, 2026

   

In this episode of CL Talk, managing editor Tim Wilbur speaks with Alexi co-founder and CEO Mark Doble about the risks law firms face as they rush to deploy AI. Doble explains why he believes traditional multi-tenant cloud tools have become “completely untenable” for serious legal work, argues that single-tenant private cloud environments are critical to protecting client data and law firm IP, and outlines how firms should divide objective, verifiable tasks from subjective, judgment-driven work when implementing AI. He also outlines practical design principles to reduce hallucinations, explains why firms need in-house champions who can challenge vendors, and warns about the reputational and financial consequences of treating AI output as work that doesn’t require review.

Inside Thomson Reuters' legal AI buy-or-build moment

Inside Thomson Reuters' legal AI buy-or-build moment

Oct 06, 2026

Why the company added a proprietary model to its existing multi-model AI strategy, with David Wong and Alexander Kardos-Nyheim

Legal AI proficiency: what Canadian lawyers keep getting wrong

Legal AI proficiency: what Canadian lawyers keep getting wrong

Sep 29, 2026

Samuel Puchala of LexisNexis Canada on prompts, hallucinations and the rise of agentic AI

Law firm culture suffers when rock star bullies run the show

Law firm culture suffers when rock star bullies run the show

Sep 22, 2026

HR lawyer Laura Williams on psychological safety, accountability, and inclusive leadership

Correctional Negligence: Holding Canada's Institutions Accountable

Correctional Negligence: Holding Canada's Institutions Accountable

Sep 17, 2026

When governments fail people in custody, who holds them to account? A leading civil litigator explains.

Fasken on agentic AI: building the tools lawyers actually use

Fasken on agentic AI: building the tools lawyers actually use

Sep 08, 2026

Clarke Barnes and Andrea Alliston on co-developing Walter AI, choosing Legora, and adoption