The Pensions Regulator publishes AI plan setting governance expectations for trustees and scheme managers

The Pensions Regulator has published an AI plan that sets initial expectations for how pension trustees and scheme managers should govern the use of artificial intelligence. The plan says AI can improve administration, decision-making and member engagement, but also brings risks including scams, bias and cyber threats. Accountability for outcomes remains with trustees and scheme managers even where activities are delegated or decisions are supported by AI. Pending more detailed guidance later in the year, schemes are expected to put clear governance around AI use, assure themselves that administrators, service providers and advisers have robust arrangements, test and monitor systems, assess and regularly review risks and controls, strengthen data quality and data protection compliance, and guard members against AI-driven fraud. The plan also sets four regulatory priorities for TPR: aligning with the Financial Conduct Authority across the pensions sector and supply chain, improving data foundations, supporting responsible innovation through its innovation service, and expanding its own use of AI and advanced analytics. In scam prevention, an AI-enabled process has assessed more than 2,000 websites and supported the removal of 29 high-risk sites. TPR plans to engage with the industry over the summer, publish fuller good practice guidance later in the year, and report annually on progress as the plan evolves.