Pension trustees urged to strengthen AI oversight as adoption grows

LWDB

Artificial intelligence is becoming increasingly common across the pensions industry, prompting calls for trustees to strengthen governance, protect member data and maintain human oversight as adoption expands.

An April 2026 industry survey found that all respondents were using AI, compared with 87% in 2025. Pension savers are also using the technology, including when preparing complaints through the Internal Dispute Resolution Procedure (IDRP).

AI offers opportunities to improve pension administration, streamline processes and support more efficient service delivery. However, trustees must ensure that its use is appropriate, properly monitored and supported by effective controls.

The Pensions Regulator (TPR) has taken a cautiously positive approach, encouraging trustees to identify specific problems that AI can solve rather than adopting technology without a clear purpose. AI should remain a supporting tool rather than replacing human decision-making, with trustees retaining responsibility even when administrators or advisers use the technology on their behalf.

Governance arrangements should be established before or alongside implementation, ensuring that risks are understood and appropriate safeguards are in place. TPR is expected to publish principles-based guidance for trustees during 2026.

Six key areas have been identified for pension administrators and advisers to address when using AI. These include understanding where AI is being used within a pension scheme, establishing who checks its outputs and whether those checks are effective.

Trustees should also understand how member data is handled, what procedures apply when AI produces incorrect results, whether systems can operate autonomously and whether providers’ insurance policies cover errors involving AI.

Written evidence and clear explanations should be obtained from service providers, with the level of scrutiny reflecting the nature and extent of AI use within each scheme.

The growing adoption of AI creates opportunities to improve pension administration, but it also increases the need for accountability and effective risk management. Trustees do not need to become technology specialists, but they must understand how AI affects their schemes and challenge providers where assurances are unclear.

Law Debenture Corporation plc (LON:LWDB) provides a wider range of services including corporate and pension trusts, process agent services, treasury management, corporate services including for special purpose vehicles, structured finance administration and whistleblowing services.

Share on:
Find more news, interviews, share price & company profile here for:

Latest Company News

Pension trustees urged to strengthen AI oversight as adoption grows

AI adoption across the pensions industry reaches 100% among surveyed respondents, increasing the focus on governance, member data protection...

Law Debenture completes £2.4bn Royal Mail pension transfer covering 100,000 members

Law Debenture has completed the transfer of £2.4bn in Royal Mail defined contribution pension assets covering approximately 100,000 members...

Six key challenges reshaping pension scheme governance

Six key challenges are reshaping pension scheme governance, from regulatory requirements and trustee succession to administration risks and Master...

UK pension reforms raise new challenges for DC schemes and employers

Law Debenture examines UK DC pension reforms, master trust governance and the inheritance tax changes affecting unused pension funds...

Law Debenture highlights opportunities to unlock value from university pension schemes

Law Debenture identifies opportunities for universities to reduce pension costs, lower employer contributions and release surplus funds from self-administered...

Law Debenture examines trustee enforcement risks in SriLankan Airlines default

The SriLankan Airlines default illustrates the challenges trustees can face when enforcing bondholder rights amid sovereign involvement, jurisdictional issues...

Search