TPR asks pension trustees to examine aggregate AI investment exposure

In October 2026, The Pensions Regulator issued guidance on AI concentration risk for pension trustees. Published on 22 October, it asks schemes to understand aggregate investment exposure and test diversification under stress, with actions proportionate to their circumstances.

Look across investment mandates

A practical trustee discussion could start by asking advisers to explain the underlying dependencies across funds and asset classes. Different labels or managers do not, by themselves, answer whether several investments rely on the same economic expectations. The analysis should identify measured holdings, estimates, incomplete information and the date of the underlying data.

Stress scenarios should then be connected to the scheme’s objectives and available responses. Committees should record who owns the assessment, what evidence would trigger a further review and when advisers will report back. A scenario should remain clearly distinguished from a predicted event.

Keep decision authority clear

For multinational employers, a common framework could help benefits teams identify recurring questions across pension arrangements. It should preserve local scheme differences and the responsibilities of trustees and advisers. Group reporting should support those decisions without substituting for them.

Member communications should explain the purpose of the review without implying that an assessment removes uncertainty. A clear board paper would distinguish evidence about current holdings from assumptions about future conditions, then record any proposed action or reason for continued monitoring.

The useful next step for benefits leaders is a structured conversation with the appropriate decision-makers. A dated exposure assessment, explicit uncertainties and assigned follow-up would make the regulator’s guidance operational while avoiding unsupported conclusions about the future performance of AI-related investments.

The review record should also identify which questions require further specialist advice before a decision.