What happened
On 29 October 2026, The Pensions Regulator (TPR) published new guidance (ref PN26-19) encouraging trustees of pension schemes to explore how AI could deliver better member outcomes while understanding and managing risks. It sets out seven principles for trustees and scheme managers (start with the problem, human responsibility remains central, governance is ongoing, plan for disruption, good AI starts with good data, transparency and explainability, and fair/accessible outcomes) and is intended to help trustees understand how their existing duties and good-governance obligations apply to AI rather than creating a separate set of regulatory requirements.
Why it matters
TPR is the UK pensions supervisor, and this guidance confirms that trustees remain legally accountable for decisions and member outcomes whether or not AI is used — including by third-party providers. It signals that UK pension-scheme governance of AI (including AI used by advisers and providers) will be scrutinised under existing duties, and it mirrors the FCA's AI guidance while being pensions-specific. Given ~76% of UK defined-benefit schemes report some use of AI by trustees, this materially sets the compliance benchmark for the sector.
Action needed
Trustees and scheme managers should map where significant AI is used (directly or via providers/advisers), ensure oversight and governance are in place, and be ready to challenge providers on AI use under the seven principles, given accountability rests with the trustee regardless of AI involvement.