AI is increasingly embedded in day-to-day work across professional services, but a new survey suggests the core challenge is not adoption. For many legal, tax, audit, and compliance teams, the issue is execution, governance, and the ability to deliver AI outputs that clients and regulators can trust.
In its Future of Professionals Report 2026, Thomson Reuters warns that a widening divide between AI ambitions and operational reality is becoming a measurable business risk. The report, based on a global survey of 1,800+ professionals, points to three connected pressures: firms may be exposing themselves to “shadow AI” risks, talent may start leaving roles where the promised value of AI does not materialize, and clients may reconsider vendor relationships when AI-enabled quality improvements fail to show up.
AI usage is widespread, but value delivery lags
Thomson Reuters frames the timing as an inflection point for the legal and accounting ecosystem. According to the survey, 74% of professionals say they use AI tools every week. However, the same respondents largely doubt their organizations are converting that usage into reliable outcomes.
The report states that 91% of professionals believe their organizations are falling short of what AI can deliver. It also highlights that, even where firms have an AI strategy, execution may not reach day-to-day workflows: 35% say their ambitions are not reflected in what they actually do, and nearly 1 in 5 report that their organization still lacks a clear strategy.
For business leaders, the implication is straightforward but costly. In high-liability fields, AI is not just another productivity tool. Outputs can influence legal decisions, regulatory filings, audit work, and client advice. If governance and quality assurance lag behind, the operational gap can become a reputational and financial one.
Shadow AI creates “invisible” risk inside organizations
A major theme in the report is the use of AI tools that may not meet internal requirements. The survey indicates that one third of lawyers, accountants, and compliance professionals say they use unsanctioned AI at work. That figure rises to 41% among professionals who believe their organizations are moving too slowly on AI.
Thomson Reuters also reports that many professionals see specific safeguards as essential. Respondents said AI must protect confidential data (96%), rely on verified and authoritative content (94%), and produce outputs that can be explained and defended (90%). Yet the report adds that 41% of respondents say they lack access to “professional-grade” AI tools that meet these standards.
This combination can produce a governance failure mode: employees seek faster or more capable tools, but firms are not equipped to provide vetted alternatives. The result is “shadow AI,” where usage may be difficult for organizations to monitor and control.
Talent risk increases when AI expectations are not met
Professional services firms also face a retention challenge tied to AI delivery. The report states that 24% of professionals who experience a gap between what AI technology can do and what their organization delivers are considering leaving within two years. Some respondents indicate even shorter timelines, with 13% considering leaving within 12 months.
At the same time, leadership perceptions may lag behind employee sentiment. The survey suggests that almost half of senior leaders believe meaningful talent pressure is at least three years away, potentially underestimating how quickly dissatisfaction can translate into departures.
Access to professional-grade AI appears to play a role in recruiting and mobility. The report notes that 62% of respondents say access to such tools would influence whether they accept a new role. Among those already using AI tools, nearly one in three say they would turn a role down without them.
Clients are demanding AI-enabled quality, then reassessing providers
Beyond internal operations, the report points to client-side expectations shifting toward measurable AI-enabled quality improvements. It says 78% of corporate clients view these improvements as very important or essential, but only 6% believe most providers are delivering them.
That gap is beginning to affect commercial relationships. Thomson Reuters reports that 32% of corporate clients plan to reconsider provider relationships within 12 months. It adds that about a third of those reconsidering would place more than $1 million in annual work at risk.
Using the survey’s parameters for the U.S. legal and accounting markets, Thomson Reuters estimates that this would translate to about $143 billion in revenue under active reconsideration within 12 months. The estimate is tied to client reassessment behavior described in the report rather than a specific forecast of losses, but it underscores how contract decisions may increasingly hinge on AI delivery capability and accountability.
Why the “accountability benchmark” matters in regulated work
Thomson Reuters argues that in professions with real liability, the bar for AI outputs must be higher than general productivity use. The report’s messaging emphasizes the need for transparency, verifiable content, and strong privacy and security controls.
While the company markets its own approach, the broader industry implication is that governance is becoming part of the value proposition for professional services technology. As clients ask whether AI can improve quality, they will likely also ask what safeguards exist, how outputs are checked, and how organizations can stand behind work product.
For firms, the survey suggests AI implementation is shifting from experimentation to operational discipline. That includes deciding where AI can be used, how outputs are validated, what data policies apply, and what escalation or human review looks like when stakes are high.
What to watch next
The report’s central message is that AI adoption alone does not resolve business risk. If firms cannot operationalize AI safely and consistently, they may face three simultaneous outcomes: more unsanctioned tool use internally, increased talent churn, and client contract reviews.
As professional services vendors compete on AI features, the market may increasingly reward organizations that can demonstrate governance and repeatable quality. That would put implementation, monitoring, and accountability closer to the center of procurement decisions, not just experimentation.
