The Future of 2026 Proxy AI: Forward-looking Report (English Edition)

Foresight paper Publication date:31March2026 This paper aims to foster debate and discussion among our stakeholders. It should not betaken as an indication of current or future policy by any of the member regulators of the This paper draws some insight from the DRCF’s Autumn 2025 public call for views on AgenticAI, and the DRCF would like to acknowledge all the respondents for their contributions. Contents Contents...........................................................................................................................3 Defining Agentic AI..............................................................................................................................4Emerging Opportunities and Risks.......................................................................................................4Ensuring Adoption of Agentic AI is Safe and Trusted..........................................................................5 1. Introduction.................................................................................................................8 2. Agentic AI: an Explainer................................................................................................9 2.1 Defining Agentic AI.........................................................................................................................92.2 The Sociotechnical System...........................................................................................................112.3 State of Development..................................................................................................................13 3.1 Emerging Opportunities...............................................................................................................163.2 Emerging Risks.............................................................................................................................18 4. Ensuring Adoption of AI Agents is Safe and Trusted: An Overview...............................21DRCF Member Regulator Remits.......................................................................................................21 5. Potential future developments and regulatory considerations.....................................26 5.1 Governance considerations.........................................................................................................275.2 Data protection and cybersecurity considerations.....................................................................305.3 Consumer Rights and Interests considerations...........................................................................32 6.1 DRCF next steps...........................................................................................................................386.2 Next steps for member regulators..............................................................................................38Contact us..........................................................................................................................................40 Executive summary This paper serves as a forward-looking exploration of agentic AI and how UK regulatory frameworkscan help realise the opportunities of this technology in a responsible and safe way. Published by theDigital Regulation Cooperation Forum (DRCF)—comprising the Competition and Markets Authority(CMA), Financial Conduct Authority (FCA), Information Commissioner’s Office (ICO), and Ofcom—thedocument aims to inform and foster stakeholder debate. The DRCF emphasises that regulation Defining Agentic AI Agentic AI represents a shift from AI as a tool, to AI as an agent that acts on behalf of users. Unlikestandard generative AI which responds to queries and creates outputs, agentic systems can assessgoals, plan workflows, and execute actions autonomouslyto impact real-world environments andinteract with people or other agents. Agentic systems are made of the following layers: (1) the coretechnical system (individual AI agents and their methods of communication), (2) the externalenvironment (other software systems, the physical world) where actions take effect, and, as Emerging Opportunities and Risks The move toward agentic AI engenders opportunities and risks for consumers and the economy. Interms of opportunities, consumers could benefit significantly, for example from agents that wouldhandle "life admin" (e.g. booking holidays, renewing policies),potentially reducing friction andimproving accessibility. Businesses could potentially realise significant productivity gains in bothcustomer engagement (front office) and internal operations (back office), such as automated Agentic AI, however, also presents risks, including the whole range of risks that apply to AI morebroadly and are covered elsewhere. Some of these risks might be increased in the case of agentic AI,and indeed there may also be new or novel risks. Delegating a task to an agent comes with thepossibility that it mig