DSIT's AI Assurance Roadmap Takes its Next Step: techUK to Lead the Industry Advisory Group of AI Assurance Consortium
9 Jun 2026 03:03 PM
The launch of the AI Assurance Consortium, chaired by BCS and supported by DSIT, marks a significant step forward in professionalising AI assurance at scale. techUK is proud to be part of the Consortium and wider roadmap as a delivery partner, leading the Industry Advisory Group and ensuring that the voice of industry, the companies building, buying and deploying AI is embedded throughout the process.
The Consortium has a clear mandate for its first year: develop a voluntary code of ethics for AI assurance practitioners; build a skills and competencies framework that maps what good looks like across different assurance roles; and map the data and information access requirements that assurance providers need from their clients. These are the building blocks for a future professional certification scheme, with AI auditing, the most mature and consequential corner of the market, as the likely starting point.
Artificial intelligence is moving fast. The systems being deployed today in financial services, healthcare, justice and beyond are making decisions that affect people's lives, livelihoods, and rights. We know how to build AI and of equal importance, we continue our work to address the market barriers of the UK's AI assurance ecosystem. We know how timportant this area is and we are committed to doing more of it.
That is what AI assurance is for. Not a compliance checkbox, but the active, evidence-based work of evaluating, measuring and communicating whether AI systems are behaving safely, fairly, and as intended. It is the infrastructure of justified trust. And right now, the UK is building that infrastructure and building it well.
The UK's AI assurance market is nascent but growing fast, worth over £1 billion in GVA in 2024, with the potential to reach nearly £19 billion by 2035 if the right conditions are created. Over 12,000 people already work across the UK’s AI assurance ecosystem. And yet the sector faces real structural challenges: a shortage of professionals with the right combination of technical, legal, ethical and sector-specific skills; no agreed competency framework; no clear career pathways; and no professional body to set and uphold standards.
The people doing this work, our membership companies’ Responsible AI leads, as well as existing AI assurance firms, auditors, evaluators are largely piecing it together themselves. techUK's April 2025 paper titled Mapping the Responsible AI Profession: Current Practice and Future Pathways, set out in detail the three critical gaps undermining RAI practitioners today: unclear role definitions, absent career pathways, and no standardised skills or training frameworks.
This is not a new concern for techUK. Our Ethics in Action: From White Paper to Workplace paper was an early attempt to bridge the gap between government's ethical principles for AI and the practical reality of how industry implements them, mapping how assurance mechanisms correspond to each of the five government-backed ethical principles and what that looks like in practice. That work, alongside our involvement in the launch of DSIT's Portfolio of AI Assurance Techniques with CDEI, reflects the depth of techUK's engagement with this agenda over several years. The consortium is not a starting point for industry, it is the much-needed next chapter.
This matters because demand is real and growing. Organisations in safety-critical, highly regulated sectors; defence, financial services, health and social care, justice, education, are increasingly asking for independent verification that their AI systems can be trusted. techUK's sector-specific AI Assurance Webinar Series explored exactly this challenge across each of these domains, and the consistent message was clear: appetite for assurance exists, but the profession to deliver it credibly does not yet. That work continues — techUK is currently working on our next paper and will be hosting a half-day conference on AI Assurance in Critical National Infrastructure, examining how assurance must evolve to meet the specific demands of sectors in the Cyber Resilience Bill like data centres, telecoms, energy, water and transport where the stakes could not be higher.
Investors are requiring assurance evidence as part of due diligence. Insurers are factoring assurance practices into coverage decisions, a dynamic techUK explored in depth in our September 2025 piece on AI insurance and a deep dive webinar. Procurement teams are beginning to treat AI governance as a contractual obligation, a theme techUK has examined closely given that most organisations don't build AI, they buy it, and that makes procurement one of the most consequential control points in responsible AI adoption.
techUK's role in the consortium reflects the breadth of that challenge. BCS leads on professionalisation bringing its experience and certification expertise.. NPL and AISI anchor the innovation and measurement work mentioned in The Trusted Third-Party Roadmap with the AI Centre for Measurement. And techUK, through the Industry Advisory Group, will connect the ecosystem: the companiesbuilding assurance tools andthe buyers deploying AI at scale. Our December 2025 report on sector-specific AI assurance applications, titled A Maturing AI Assurance Ecosystem: Sector Specific Applications, demonstrated the range and depth of implementation insight that industry brings to this work, insight that must be embedded across all workstreams of the roadmap, not just a single stream.
The UK has a real opportunity here. Relative to the size of our economy, our AI assurance market already outpaces the US, Germany and France. We have world-class professional services expertise, strong regulatory institutions, and a government commitment (reflected in the roadmap) to get this right.
techUK has been part of building that foundation, and we are committed to ensuring the consortium delivers on its promise.
The building blocks are in place. The partnerships are formed. Now comes the work.