Thomson Reuters' data on AI adoption across law firms makes for seductive reading. 41% using generative AI, 47% of corporate legal departments—these are real jumps from 2025 baseline figures. For UK mid-market practices, the message is clear: AI is no longer optional, it's table stakes. But there's a dangerous gap between 'using AI tools' and 'implementing AI responsibly under the SRA Code and FCA Consumer Duty principles'. Most of the firms in that 41% are likely running ChatGPT, Harvey, or Legora without proper audit trails, without clear accountability for outputs, and without documented governance frameworks. The regulators know this. They're watching. And they will ask.

What this story reveals is an industry in a race, not a journey. Law firms, insurers, and accountancy practices are adopting point solutions—LLMs wrapped in user-friendly interfaces—at pace precisely because they can. There's no friction. But friction exists for a reason. The EU AI Act is weeks away from enforcement. The ICO is tightening guidance on generative AI and UK GDPR compliance. The SRA has made clear that firms remain liable for AI outputs under their existing codes. Yet the adoption curve shows no sign of flattening or becoming more thoughtful. Speed is winning. Responsibility is trailing.

Here's Trovix's honest assessment: most current AI adoption in professional services is point-solution thinking dressed up as transformation. Harvey looks impressive on document review. Luminance excels at pattern spotting. But neither tool exists within a compliance ecosystem. Neither tool generates the audit evidence that meets FCA Consumer Duty PS22/9 requirements. Neither tells you clearly who is accountable when the AI gets it wrong. The firms winning right now are not those with the shiniest LLM—they're those implementing AI through a governance layer first, then choosing tools second. That means Trovix Audit as a prerequisite, not an afterthought. It means Trovix Aria delivering RAG-based knowledge assistance with accountability baked in, not bolted on. It means understanding that 'using AI' and 'governing AI responsibly' are not two separate projects—they're one project done wrong, or one project done right.

If you run a mid-market legal, insurance, financial services or accountancy firm in the UK, the question is not whether to adopt AI—you already know you must. The question is whether you will be the firm that explains to the regulator why you chose speed over governance, or the firm that chose sustainable implementation. Start now with two things: first, audit your current AI usage honestly (you probably have more than you think); second, establish who owns accountability for AI outputs in your business. Then—and only then—add new tools deliberately into a framework that someone senior can explain to a regulator without embarrassment. The firms in Thomson Reuters' 41% figure who move fastest on governance will be the ones regulators cite as examples of good practice, not as warnings.

Source: Bloomberg Law