Building on the DRCF AI & Digital Hub pilot, which demonstrated the value of informal, cross regulator collaboration, the DRCF launched the Thematic Innovation Hub, beginning with a focus on Agentic AI. This thematic approach enables regulators to better understand emerging risks and opportunities and to engage earlier with innovators developing complex new technologies.
Kate Jones, DRCF CEO, reflections:
The Thematic Innovation Hub was launched with a precise aim in mind - to serve as a bridge between innovation and regulation, and to ensure we maintain and nurture an active dialogue between industry and regulators.
That’s exactly what the Hub has achieved since its launch in October 2025, building on the success of the DRCF AI & Digital Hub pilot. We kicked off our first series of engagement on a topical theme, which has recently shaken the whole digital ecosystem: agentic AI (AI systems capable of autonomous decision-making and initiating actions without direct human prompts).
With every novel technology, it’s fundamental to ensure regulators are tuned in on challenges and opportunities industry are facing, and that innovators are able to share first hand insights on emerging technologies. We ran a call for input and have hosted a range of engagement sessions to facilitate dialogue, including a webinar, policy sessions and two roundtables in London and Edinburgh.
Agentic AI brings practical, high-level challenges that are showing up fast, and growing as AI becomes increasingly embedded in our everyday life. Agents operate at machine pace while assurance runs at human speed, liability becomes blurry as agents make autonomous decisions, and user literacy becomes crucial to ensure consumers understand risks. But it also comes with exciting opportunities – it can provide an incredible productivity boost for employees, empower consumers by providing personalised and real-time assistance, enable smaller firms to do more with less, unlock new levels of knowledge and so much more.
We’ve taken stock of lessons learned to make sure our findings can provide a helpful injection in the policy debate around AI, and that regulators’ policy-making is informed and supported by industry feedback. We gathered some key takeaways as part of our engagement with firms:
1. Speed is outpacing assurance:
Agentic systems act and iterate far faster than current governance processes can review. A solution to close the gap could be releasing capabilities in steps and monitoring them in real time.
2. Cross-regulatory coherence:
Firms asked for cross‑regulator coherence and common standards across the end‑to‑end user journey. They also stressed interoperability to support competition and better consumer outcomes, and practical cross‑border compatibility to align to existing requirements like the GDPR and EU AI Act and regulators’ existing international networks.
3. Delegation and consent:
As agents keep working after an initial prompt, authority and intent can blur over time; patterns like intent‑expiry (e.g., a mortgage‑finding task ending when the mortgage is secured) and prominent, in‑journey consent renewal help keep users in control.
4. Platform concentration creates power and resilience risks:
Heavy reliance on a few model providers strengthens asymmetries and can introduce failures; industry is calling for interoperability, and regulators’ attention to market dynamics so smaller players aren’t locked out.
5. Regulate for outcomes along an autonomy spectrum:
Firms shared that regulation of agentic AI should not come with a blunt, binary label (‘agentic/not agentic’). Instead, regulators should look at what the system actually does and how much autonomy it has, then set duties that fit the risk in that specific context (e.g., B2B vs B2C, finance vs general productivity). In other words, regulate for outcomes, calibrated along an autonomy spectrum rather than a one-size-fits all rulebook.
6. Liability needs tiers - and a clear ‘issuer of record’:
With third party agents acting inside enterprise stacks, responsibility fragments; industry pointed to KnowYour-Agent checks, tiered accountability, and even ‘smartcontract’ rails to track duties and enforce killswitches when things go wrong.
7. Trust is built through literacy and visible guardrails:
User education, standardised warnings/T&Cs, and cross-regulatory sandboxes can support literacy and consumer appetite for interoperable agents and agentic systems.
8. Data protection considerations:
Data Protection law is technology-neutral, but as with all new technologies, each application is different: how agents and LLMs uphold information rights can look very different in different sectors and use cases. Firms shared that UK GDPR and privacy requirements can be time-consuming to navigate, with smaller organisations, startups, and sole traders in particular needing clarity and guidance from regulators.
We conducted this first Thematic Hub series on Agentic AI in parallel to our Horizon Scanning research on Agentic AI. For a more detailed exploration of potential regulatory implications of Agentic AI, please refer to our latest piece: The future of Agentic AI report.
Participants reflections:
Sarwar Khan, Director of Government Affairs, Salesforce UK & Ireland said:
“Participating in the DRCF’s Agentic AI workstream has been valuable for Salesforce to help shape the future of the UK’s digital economy. As we move from assistive tools to autonomous agents, the proactive and collaborative approach taken by the DRCF is exactly what is needed to build public trust and certainty.
For Salesforce, this engagement is critical; it ensures that the UK remains a global leader in innovation while upholding the highest standards of AI safety and ethics. We believe a pro-innovation, interoperable framework, as championed in this series, is the key to helping our customers deploy agentic AI with confidence, ultimately driving productivity, growth and transformation priorities across the UK public and private sectors.”
Kirsty Ireland, Senior Director, Global Public Affairs, Skyscanner said:
“We were pleased to take part in the DRCF’s industry roundtable on agentic AI, one of the most exciting and fast-evolving areas for both industry and consumers. At Skyscanner, we are exploring how agentic AI can enable more seamless travel search and booking experiences for our users.
The discussion highlighted the importance of maintaining a competitive, fair and open ecosystem, something we strongly support, while also touching on important themes around clear consent and well-defined delegation frameworks that give businesses the confidence to innovate and move faster. "
The insights shared by both regulators and industry were invaluable, and we believe continued open dialogue will be essential as these technologies develop. We support the UK government’s ambition to position the UK as a global leader in AI innovation, underpinned by a regulatory environment that enables businesses to innovate and thrive.”
Mohamed Zamzam, Co-Founder & CRO, compostable.ai said:
“Compostable AI was glad to contribute to the DRCF's Thematic Innovation Hub on Agentic AI. What gives us confidence isn't just that regulators are paying attention — it's that they're asking the right questions. Outcome-based regulation calibrated to an autonomy spectrum is the right foundation. If the UK can move from dialogue to coherent, interoperable standards at pace, it has a genuine opportunity to lead the world in trustworthy agentic AI deployment."
