Trust is the most valuable asset, the Pensions Regulator has said as it publishes its artificial intelligence plan and tells schemes to ensure safe and responsible adoption “in members' interests”. Detailed guidance is set to be published later in the year, with a call for industry input over the summer.

The plan, published on Wednesday, describes TPR’s approach to supporting the use of AI in pensions and clarifies its expectations of trustees, administrators and scheme managers. 

In addition, it sets out steps the regulator says it will take to enable safe AI adoption and AI-powered innovation in the pensions sector, and describes how TPR itself will use the new technology. 

It describes the key benefits of AI in pensions to lie in trustee-led communications and guidance, as well as administration, while TPR sees risks in people using chatbots for financial advice and warns AI may be turbo-charging a range of scam activities. The regulator is urging trustees, administrators and scheme managers to act now to support safe adoption, enable innovation and to protect members from AI-driven fraud. 

TPR itself will continue to use AI and advanced analytics to help it identify risks and target scams, including the analysis and takedown of high‑risk scam websites, it said.  

“AI has the potential to transform pensions for the better: improving how schemes are run, how members are supported, and how the system as a whole delivers value. But trust is the most valuable asset in our system, and that trust depends on the safe and responsible adoption of AI in members' interests,” said chief executive Nausicaa Delfas.  

“Our message to trustees, administrators and scheme managers is clear: act now. Put strong governance in place, invest in data quality, understand where and how AI is being used in your scheme, and protect your members from AI-driven fraud,” she said.  

TPR stressed that trustees and scheme managers remain accountable for outcomes even when activities are delegated. This means they must: establish clear governance for AI use, and make sure administrators, service providers and advisers have similarly robust arrangements; carry out testing, assurance and monitoring of AI systems; identify and evaluate risks, with appropriate controls reviewed regularly; and prevent members being scammed by being aware of AI-driven fraud methods and responding effectively. 

They also need to have a clear data strategy, ensure scheme and member data is of high quality, and comply with data protection legislation, including in automated decision-making. In addition, TPR expects that trustees seek advice when considering or implementing innovations.  

TPR’s focus areas will be to ensure schemes are well governed, engaging with schemes on data, fostering innovation and using AI to become more effective. The regulator said it will report on progress annually. 
 

The pensions industry is already adopting AI, and Helen Forrest Hall, chief strategy officer at the Pensions Management Institute, said it will become an integral part of pensions.

 

“AI brings real opportunities to improve outcomes, but it also introduces new risks. Strong governance, high‑quality data and robust controls will be essential to maintaining trust as schemes embrace new technologies," she said.

There have already been numerous court cases where mainly law firms have been found wanting for relying on AI without appropriate checks, with the technology ‘hallucinating’ – for example, making up precedent cases. The Financial Times reported recently that big four advisory EY had to retract a study on supermarket loyalty schemes as it contained hallucinations and fake footnotes.

Earlier this year, the Financial Conduct Authority ran a call for input to a review of the impact advanced artificial intelligence will have on consumers, retail financial markets and regulators, led by Sheldon Mills. Just before it was launched, MPs warned that while AI can bring benefits to consumers, the UK is unprepared for AI-related shocks, exposing the public and the financial system to potentially serious harm.