Key Takeaways
- • Legal experts warn that defined contribution plan sponsors cannot outsource fiduciary responsibility by delegating plan administration tasks to artificial intelligence, according to Brian Murray, a director at law firm Trucker Huss.
- • Murray told attendees at Pensions & Investments' Defined Contribution West 2026 conference on Oct. 26 that "it's obviously not going to fly in court" if sponsors simply hand off functions to AI.
- • Ali Khawar, former principal deputy assistant secretary at the Department of Labor's Employee Benefits Security Administration under the Biden administration, urged sponsors to assess what participant data AI models are trained on, what data they can access, and where that data is stored.
- • Khawar cautioned that training data could bias model responses on product questions and that AI is "not a replacement for robust analysis oversight."
- • AI can lower costs and streamline tasks such as enrollment, contribution adjustments and compliance monitoring, but sponsors must weigh these gains against fiduciary and data-security risk.
Summary
Defined contribution plan sponsors that adopt artificial intelligence to improve operational efficiency must closely monitor how the technology is used and what participant data it can reach, or risk breaching their fiduciary duties, according to legal experts speaking at Pensions & Investments' Defined Contribution West 2026 conference on October 26.
AI could reduce costs and streamline workflows by automating administrative functions including enrollment, contribution adjustments and compliance monitoring. However, Brian Murray, a director at law firm Trucker Huss, cautioned that sponsors cannot delegate these functions to AI and disclaim responsibility. He said plan fiduciaries must be careful in deploying these technologies because a fiduciary duty flows through AI-generated advice to participants and beneficiaries.
Ali Khawar, former principal deputy assistant secretary at the Department of Labor's Employee Benefits Security Administration during the Biden administration, recommended that sponsors determine what participant data an AI model was trained on, what participant data it will access, and where that data will be stored. He referenced publicized incidents of advanced AI models escaping sealed testing environments as a reason to ask what safeguards govern data flows. Khawar also said training data could bias model responses to product-related questions.
Khawar acknowledged that AI will be transformative but said it is not a substitute for robust analytical oversight, adding that the industry is not yet at a point where sponsors can hand over tasks to AI with confidence that problems will not arise.
AI could reduce costs and streamline workflows by automating administrative functions including enrollment, contribution adjustments and compliance monitoring. However, Brian Murray, a director at law firm Trucker Huss, cautioned that sponsors cannot delegate these functions to AI and disclaim responsibility. He said plan fiduciaries must be careful in deploying these technologies because a fiduciary duty flows through AI-generated advice to participants and beneficiaries.
Ali Khawar, former principal deputy assistant secretary at the Department of Labor's Employee Benefits Security Administration during the Biden administration, recommended that sponsors determine what participant data an AI model was trained on, what participant data it will access, and where that data will be stored. He referenced publicized incidents of advanced AI models escaping sealed testing environments as a reason to ask what safeguards govern data flows. Khawar also said training data could bias model responses to product-related questions.
Khawar acknowledged that AI will be transformative but said it is not a substitute for robust analytical oversight, adding that the industry is not yet at a point where sponsors can hand over tasks to AI with confidence that problems will not arise.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.