Thomson Reuters CEO: gap widens between firms operationalising AI and those falling behind

Thomson Reuters announced its 2026 Future of Professionals report on 22 June 2026. In it, CEO Steve Hasker argues that AI is 'ready' but many firms are not, and warns that falling behind on AI implementation creates tangible risks for talent, clients and financial performance. According to Thomson Reuters, a gap is emerging between organisations that demonstrably operationalise AI and those that do not. For legal leaders, the question therefore shifts from 'should we use AI?' to 'can we demonstrate where AI adds value, where it stops and where human responsibility begins?'.

That calls for a formal AI strategy, governance, performance measurement and workflows that are auditable. The practical consequence: firms must move from loose experiments to verifiable and responsible AI use, with clear boundaries between what AI supports and what a lawyer decides.

What exactly is Thomson Reuters warning about regarding AI in law firms?

In the press release accompanying the 2026 Future of Professionals report, Thomson Reuters CEO Steve Hasker argues that the technology is ready, but that many organisations have not yet got its effective implementation in order. The core of his message is not that AI takes over the work, but that the failure to implement it thoughtfully has itself become a risk to talent, clients and financial results.

This observation does not stand alone. The earlier report Future of Professionals 2025: Mind the gap from the Thomson Reuters Institute already drew a connection between formal AI strategies and higher revenue growth, and between AI use and time freed up for professionals. The executive summary of the 2025 Generative AI in Professional Services report also describes how the adoption of generative AI in the legal sector nearly doubled in a single year, while almost half of firms still had no formal policy around AI.

Editorial analysis: this is where the actual fault line lies: the difference no longer rests in whether lawyers use AI, but in whether that use is steered, bounded and accountable. That is a governance question, and therefore a question about AI in professional legal practice.

How is AI changing the daily legal workflow?

The shift is substantive, not merely quantitative. According to the summary of the 2026 AI in Professional Services report for legal teams, the use of generative AI rose sharply again, and the share of sector-specific, professional AI solutions grew in particular. Lawyers are thereby moving away from generic tools towards domain-specific applications.

Concretely, these tasks are most affected:

• Legal research: finding and summarising sources more quickly, with the risk of incorrect or non-existent references.
• Summarising and analysing contracts and case files.
• Knowledge management: unlocking internal documentation and precedents.
• Drafting and editing work, where the lawyer checks the final text.

The lawyer's role shifts towards review, framing and supervision. This creates new points of attention around source reliability, professional secrecy and accountability. Building a defensible workflow for legal research with AI therefore means building in verification, logging and visible boundaries on what AI may decide, rather than adding them afterwards.

How do you make AI use in a legal practice demonstrably responsible?

The opinion piece How legal leaders are winning with AI in 2026 from Thomson Reuters notes that while use is rising, many firms cannot yet demonstrate that their AI investments are working. The emphasis there is on performance metrics, governance structures, cross-functional ownership and defining what should never be fully left to machines.

Editorial analysis: those seeking to close the execution gap can apply a number of design principles. This is our elaboration of the direction the reports mentioned point towards, not a literal source citation:

1. Choose professional, sector-specific AI tools and record why.
2. Define decision boundaries between human and machine for each type of task.
3. Measure the effect on quality as well as turnaround time, not just speed.
4. Make workflows auditable: which model, which source, which human check.
5. Assign ownership so it is clear who is accountable for an AI-supported decision.

That last point connects to the broader discussion about who is accountable for AI decisions and to the idea that human oversight as a design requirement for AI should be built in, not applied as a signature afterwards.

What role can a verification layer such as Vera play?

The Thomson Reuters reports describe the problem, not one specific solution. What is lacking, according to the sources, is the visible, demonstrable side of AI use. On that point, a verification layer can be supportive.

Vera is not a chatbot and not its own language model, but a privacy-focused verification layer for work involving confidential information. Vera can route a task through selected independent AI models and make verification steps, corrections, disagreements and sources visible for inspection. This supports review and gives more insight into the reasoning, but does not remove the need for human checking and does not eliminate errors. The Semantic Privacy Shield can replace sensitive document values with synthetic, session-only equivalents on EU infrastructure before AI processing; the workflow is fail-closed, so that when a privacy check fails, nothing is sent onward. The final professional judgement always remains with the lawyer.

This is how Hasker's warning translates into something concrete: not new standards, but a layer that makes visible, for each AI-supported workflow, which models were used, which checks applied and which output was reviewed by a human — usable as evidence for clients, regulators and the firm's own board.

Sources: The article relies on Thomson Reuters' 2026 Future of Professionals press release featuring statements by CEO Steve Hasker and on additional Thomson Reuters Institute reports on AI in professional and legal practice.

← All articles in this topic ← All articles