Your members are using AI. Your policy assumes they read it.

rC89: every barrister with management responsibility, the head of chambers, the management committee carries a personal duty to ensure chambers is administered competently, and the May 2026 guidance addresses chambers-level AI policies, shared systems, and records through exactly that rule. In practice, the committee has delegated the discharge of that duty to its director or senior clerk. This engagement is how it gets discharged with evidence that it was.

Then you're where most chambers are and that's exactly the position this addresses. The harder question is whether the policy matches what members actually do. A policy nobody has read, while members use AI anyway, doesn't reduce the exposure: it documents that the risk was identified and then left unmanaged. The guidance asks chambers to demonstrate a practice, and a policy describes what should happen, not what does.

Members remain individually responsible under the Handbook, and chambers has no authority over the tools they choose. But limited authority is exactly why the enquiry is the whole game: chambers can't control what members use, but it can prove it asked, recorded the answers, provided approved and/or recommended vetted options, and trained anyone who showed up. For a set of self-employed practitioners, that is what a demonstrable approach looks like and it protects chambers, and its director, regardless of the members who don't engage.

The guidance itself is non-mandatory, but it interprets duties that aren't: Core Duty 10, rC86 on outsourcing, rC87/rC89 on managing your practice and chambers, Core Duty 6 on confidentiality. It is also the yardstick a court, the BSB, or a panel client will reach for after an incident; and Ayinde and Munir show that incidents are already reaching the courts. Guidance defines what “reasonable” looks like, which is precisely why acting after it exists is cheap and explaining inaction after an incident is not.

Members may not have engaged with your AI policy because it offered them nothing. This is designed the other way round: members get a governed, best practiced-driven way to use AI that makes them faster on research, first drafts, and summarising, with clear lines on what never leaves chambers. In our experience the session sells itself once the first silk asks a practical question. The director gets the attendance record either way.

No. This is a lightweight, ISO/IEC 42001-aligned governance baseline, the standard the BSB guidance itself points to without the weight or cost of a certification programme. It's the right foundation if chambers ever pursues certification, but the point of it is to let you answer the committee, a client, or the regulator now.

Some sets can, and if your practice management team is actively maintaining a tool register, AI risk register and its associated risk treatment plan, and audit logs, you don't need us. Where we earn our place is doing the enquiry and the build quickly, in the form the guidance expects, without consuming the director's year. We bring ISO 42001 grounding the guidance references.