The Digital Regulation Cooperation Forum has published a paper exploring use of agentic AI and how UK regulatory frameworks can help realise its opportunities in a safe way. The paper looks at:
- defining agentic AI – noting it represents a shift from AI as a tool to AI as an agent, so it is no longer just responding to queries but is carrying out tasks autonomously. At the moment, most agents do however require close supervision and handle only a limited number of steps;
- emerging opportunities and risks – looking at how consumers could benefit from agents that handle life admin, and businesses could see economies in both front and back office operations. But there is the risk that the agent might make mistakes that have severe consequences and risk becoming black boxes – and the resulting lack of transparency could lead to problems in complying with consumer, contract and data protection laws. There may also be a risk of “vendor lock” where switching providers becomes difficult;
- the importance of ensuring that adoption of agentic AI is safe and trusted – each of the 4 regulators involved agrees that AI agents do not fall outside existing regimes. The FCA for example expects firms using agentic AI (or any tool) to comply with the Consumer Duty and deliver good outcomes whatever method they use;
- potential future developments and regulatory considerations – looking at the cross-regulatory implications on governance, data protection and cybersecurity, consumer rights and interests and market dynamics and competition.