“We’re Not Using AI Yet” Is No Longer a Safe Answer

For the past two years, the legal industry’s AI conversation has been dominated by cautionary tales: the sanctioned associate, the fabricated citation, the embarrassed partner explaining to a judge why five cases in a brief don’t actually exist. That narrative has done real work. It’s made lawyers appropriately wary of pasting unverified AI output straight into a filing.

But a recent piece in the Telegraph points at a shift that every managing partner, GC, and legal administrator needs to sit with: the professional risk is no longer only about misusing AI. Increasingly, it’s about not using it at all — or using it without the competence to supervise it properly. The same duty of competence that regulators have used to sanction lawyers for AI mistakes is starting to cut the other way, treating a failure to responsibly adopt available technology as its own form of professional shortfall.

The Duty Was Never About the Tool

This isn’t as radical a pivot as it sounds. Regulators in the UK and bar associations in the US have been consistent on one point: the rules of professional conduct don’t change because a machine did some of the work. The UK’s SRA has said plainly that solicitors remain personally responsible for every piece of work product regardless of whether AI touched it, and its guidance ties this back to the existing duty to maintain competence and keep skills current — not a new AI-specific rule, just the old rule applied to a new tool. The ABA’s Formal Opinion 512 does the same thing in the US, requiring lawyers to have “a reasonable understanding of the capabilities and limitations” of any AI tool they use.

Here’s the part that’s easy to miss: a duty to understand a tool’s capabilities and limitations is very hard to satisfy from a standing start, under deadline pressure, with no training. Firms that have spent the last two years avoiding AI because “it’s risky” are now sitting on a different kind of risk – a competence gap that shows up the moment a client, an opposing counsel, or a judge asks a direct question about how a document was prepared.

Competence Is a Team Sport

This is where the conversation has to widen beyond the partners doing the drafting. The paralegal running the first-pass document review, the legal assistant summarizing a deposition, the associate using AI to build a research memo, the intake receptionist fielding a client’s question about whether “that ChatGPT thing” is being used on their file. Every one of those roles now touches the firm’s AI competence posture. A firm can have a brilliant AI policy sitting in a binder and still be exposed if the people executing day-to-day workflows were never actually trained on it.

That’s the gap we see constantly at Savvy Training & Consulting. Firms buy the tools like Copilot, Claude, purpose-built legal AI, and assume adoption equals competence. It doesn’t. Competence is built through structured, role-specific training: what a paralegal needs to know about verification is different from what a partner needs to know about client disclosure obligations, which is different again from what an admin needs to know about confidentiality and data handling.

Building AI Competence Without Building It From Scratch

This is exactly why we built the SavvySMART Content Library and SavvyAcademy ProPlus LMS – a ready-to-deploy foundation of technology training content that firms can roll out immediately, rather than trying to reinvent AI literacy training internally while also running a practice. It covers the full spectrum: AI fundamentals and prompt literacy for administrative staff, verification and supervision protocols for associates and paralegals, and risk, disclosure, and governance considerations for partners and practice group leaders, plus the security awareness layer (via our KnowBe4 partnership) that has to sit underneath all of it.

For firms that want it hands-on, we also deliver on-site, remote, and hybrid training with curricula custom-built around your actual technology stack, not a generic AI 101 deck, but a program mapped to the tools your people use every day and the matters they actually work on.

The regulators have been consistent: the duty of competence doesn’t bend for new technology, it absorbs it. The firms that treat AI training as infrastructure, not an afterthought, are the ones who’ll be able to answer “how do you ensure your AI use is competent and supervised?” with confidence instead of a shrug.