What the BCS AI assurance consortium actually launched

The AI assurance profession UK conversation changed on 8 June 2026. At its AI Adoption Summit, the Department for Science, Innovation and Technology sized the domestic assurance market at £1.01 billion in gross value added today, with a projected rise to £18.8 billion by 2035.

Kanishka Narayan MP, Minister for AI and Online Safety, led the announcement. Alongside the figures came something more structural: the AI Assurance Stakeholder Consortium, convened by BCS, The Chartered Institute for IT.

This is the moment a market stops being a collection of vendors and starts becoming a profession. The full announcement sits on the GOV.UK newsroom for anyone who wants the primary source.

If you're tracking the wider picture, our UK AI assurance hub gathers the market sizing, the regulatory backdrop, and the sector applications in one place.

Who is in the room

The consortium brings together the bodies that make professions real. The UK Accreditation Service, the British Standards Institution, and the National Physical Laboratory cover accreditation, standards, and measurement.

The Ada Lovelace Institute and the Chartered Quality Institute add public-interest research and quality discipline. Independent experts including Adam Leon Smith FBCS and Professor Dame Wendy Hall round out the group.

That combination of UKAS, BSI, and NPL AI standards expertise matters. When measurement, standards, and accreditation bodies sit at the same table, the output tends to become the reference everyone else is measured against.

The four deliverables, and what they signal

Over an initial one-year period the consortium will produce four things. Each one is a building block of a recognised profession.

First, a voluntary professional code of ethics. Second, a skills and competencies framework. Third, an information-access map for assurance providers. Fourth, cross-sector collaboration to keep the work joined up.

Read together, these are the same four moves accountancy and engineering made on their way to becoming chartered professions: agree the ethics, define the competencies, set the standards, and accredit against them.

The skills framework is the part I'd watch most closely. It describes what a competent assurance practitioner must be able to do, and by extension what the tooling under them must be able to prove.

Why "voluntary" rarely stays voluntary

A code badged as voluntary often reads as optional. History suggests otherwise.

Once a government publicly sizes a market and names the standards bodies, the voluntary baseline tends to become the de facto expectation. Procurement teams adopt it, insurers reference it, and regulated buyers start asking why a supplier sits outside it.

For anyone thinking about AI assurance certification UK frameworks, the direction of travel is clear. The smart move is to operate as if the code already applies, because the gap between voluntary and expected usually closes faster than people plan for.

The maturity gap nobody likes to name

Here is the quieter problem a profession will expose. Most enterprises have an AI policy. Far fewer can show, with evidence, what a live system actually did in production.

A document describing intended behaviour is not the same as proof of actual behaviour. That gap is where PowerPoint Governance lives, and a professionalising market is precisely the thing that makes it visible.

Codes and competencies describe what good looks like. The operational capability to test a model, enforce policy on it, and continuously evidence its behaviour is what actually meets that bar. You can read how we structure that work across the AI assurance lifecycle.

What this means for the tools doing the work

A skills framework raises the evidence bar for people and platforms alike. If a practitioner is expected to demonstrate ongoing control of an AI system, the platform beneath them has to generate the proof on demand.

That is the work we built Disseqt to do. One unified platform that handles testing, monitoring, policy enforcement, audit, and AI compliance, so the evidence a recognised profession will expect exists as a matter of routine, not a fire drill before an audit.

Test & Detect. Protect & Enforce. Prove & Comply. Those three motions map directly onto what a competency framework asks for, and they run continuously rather than as a one-off snapshot. The mechanics live on our AI governance platform page.

Bottom Line

The UK has done something deliberate. It has named the market, put a number on it, and assembled the bodies that turn a trade into a profession.

Codes and competencies will define what good assurance looks like. Producing the continuous, audit-ready evidence that meets that definition is a separate capability, and it's the one worth building now rather than when the voluntary baseline quietly becomes the expected one.