Our Take
The real problem isn't that AI adoption is slow; it's that leadership is asleep while staff either leave or go rogue.
Why it matters
Professional firms face a dual crisis: talent hemorrhage and unauthorized tool use that creates security and liability exposure. The gap between staff expectation and C-suite action is widening, not closing.
Do this week
Legal/finance leaders: Audit which tools your team is actually using this week, then map gaps between approved tech and what staff need to do their jobs faster.
One-third of lawyers use shadow AI despite risks
Thomson Reuters surveyed 1,816 professionals across law, tax, accounting, compliance, and risk in 62 countries (March-April 2026). The findings are stark: 33% of lawyers, accountants, and compliance professionals are using AI tools that fall outside their organization's approved tech stack, meaning no official security controls or data governance.
The shadow AI problem worsens when organizations move slowly on adoption. Among professionals who report their firm is moving too slowly, the rate climbs to 41% using unapproved tools (per Thomson Reuters survey data). These tools lack proper security and data barriers, exposing client information and work product to unmanaged risk.
Separately, 25% of respondents said they want to leave their job within two years because their firm isn't providing adequate AI tools (company-reported). This signals attrition driven not by AI itself but by friction around access to it.
C-suite perception lags reality. Nearly half of senior leaders surveyed believe meaningful talent pressure from AI capability gaps is still at least three years away. This timeline mismatch suggests leadership does not see the present-day exit problem or believes it will correct itself.
Slow adoption creates immediate business jeopardy
The mechanics are simple: staff know AI can automate routine work. When firms don't provide approved tools, employees either leave for competitors with better offerings or adopt unauthorized software. Neither outcome is acceptable.
The shadow AI route is worse. A lawyer using an unapproved LLM for client work without encryption, access controls, or audit trails introduces legal liability, data leakage, and regulatory exposure. In regulated industries (law, accounting, finance), "almost right" outputs can trigger malpractice claims or compliance violations.
The talent route is slower but cumulative. Losing a quarter of your workforce over 24 months isn't a hiring problem; it's a strategy problem. Firms that fail to adopt approved, vetted AI tools while competitors do will face both capability gaps and brain drain.
Artificial Lawyer's reading of the data: three more years of inaction doesn't buy safety. It guarantees obsolescence.
What firms need to do now
First: Identify what your team is actually using. Run a shadow IT audit focused specifically on AI tools. Don't assume you know.
Second: Provision approved alternatives. Whether you build, buy, or partner (Thomson Reuters positions its CoCounsel product as "Fiduciary-Grade AI"), the tool must be secure, documented, and tied to training. The specific vendor matters less than the speed of rollout and clarity on liability.
Third: Tie AI access to career outcomes. If associates see AI adoption as part of firm growth and their own advancement, retention stabilizes. If it looks like a threat or a gap in your offering, they leave.
The survey covers firms of all sizes. Big Law tends to move faster and has legal resources to manage approval frameworks. In-house and mid-market firms often lag. For those, the cost of inaction is now measurable in quarterly attrition, not speculative risk three years out.