SPP AI governance framework sets risk rules for pension schemes

The SPP AI governance framework has been launched to help pension scheme trustees, advisers and administrators apply artificial intelligence with tighter controls over risk, data and oversight. The Society of Pension Professionals said the guidance is intended to support responsible AI leadership across the pensions industry while keeping trustees’ core fiduciary duties unchanged.

The framework comes as AI use becomes more common in pensions administration and advisory work, according to the SPP. For schemes handling member records, benefit calculations and communications, the practical effect is a greater focus on who validates AI outputs, what information can be entered into AI systems and how suppliers disclose their own use of the technology.

What is the SPP AI governance framework?

The framework is a practical guide for pension schemes on the governance of AI tools. It sets out five principles covering risk classification, human review, data protection, supplier controls and member communications.

The SPP said schemes should classify AI use by risk, with low, medium and high tiers. Uses classed as high risk include AI outputs that directly influence member benefits or advice, and the framework says those cases require stronger governance and robust validation.

A second principle covers meaningful human oversight. The SPP said automated decision-making should involve qualified reviewers with the authority to assess and change outcomes, rather than human checks that exist only as a formal step in the process.

Data security and privacy form another part of the guidance. The SPP said confidential scheme information and sensitive personal data should not be entered into unapproved AI systems or public AI models, reflecting the risk that pension data could be exposed outside controlled environments.

How does the guide address suppliers and members?

The framework also addresses third-party and adviser governance. The SPP said supplier contracts should be updated to require transparency on AI tools, policies for human review and protections against AI-enabled cyber fraud or deepfakes.

For members, the guide calls for clear scheme communications. The SPP said pension schemes should help prevent members from relying on inaccurate or hallucinated information generated by public AI tools, particularly where that information could affect retirement decisions or expectations about benefits.

Jo Fellowes, chair of the SPP administration committee, said artificial intelligence has shifted from an emerging technology to an everyday feature of pensions work. “The challenge is therefore not whether AI should be used, but how it can be used safely, transparently and with appropriate oversight. This SPP guide should help schemes achieve this,” Fellowes said.

The framework is available through the SPP’s guides and reports section. The organisation presented it as a practical tool for schemes seeking to adopt AI without weakening governance over member outcomes, scheme data and third-party services.

This story draws on original reporting from Finextra Research.