Key Takeaways
- Artificial intelligence (AI) supports retirement plan administration through automation, document review and data analysis, improving efficiency and participant support.
- AI outputs require validation. Plan sponsors retain fiduciary responsibility for compliance, oversight and decisions despite increased use of automation.
- Data privacy, cybersecurity and vendor oversight remain critical. Plan sponsors should assess AI controls, data protection and Service Organization Control (SOC) reporting.
Artificial Intelligence (AI) is no longer a future concept for plan sponsors. From recordkeepers and third-party administrators to internal administrative processes, AI is actively shaping the future of retirement plan systems. In many cases, plan sponsors may have limited control over its use, especially by sponsor vendors. However, as a plan sponsor, it’s important that you’re aware of how and where AI is used, its associated risks and how to effectively oversee it.
AI Use in Retirement Plans
AI is increasingly being used in retirement plan services in both obvious and less apparent ways. For participants, chatbots and virtual assistants can provide 24/7 support and answer plan-related questions. Some vendors use AI tools to offer personalized guidance based on individual goals, savings habits and risk tolerance.
From an administrative and plan sponsor perspective, AI can support several functions designed to improve the efficiency of plan operations, helping you summarize plan documents and understand complex plan provisions, data and reports. This includes breaking down census data, contribution reports and compliance testing results into clear and concise summaries.
AI can help automate routine tasks such as enrollment, loans and withdrawals while also monitoring plan activity. It can write participant communications, reminders and general plan updates. AI can screen vendor reports and other items to help flag any inconsistencies, outliers or potential fraud risks before issues arise.
Benefits of AI
When used appropriately, AI may improve speed, consistency and efficiency in plan administration. Tasks that once took hours may be completed more quickly, giving administrators faster access to information and providing participants more timely support. However, these benefits depend on reliable inputs, reviewed outputs and proper oversight.
AI Risks
Despite the many benefits, AI introduces real risk. It can produce inaccurate information but display it with confidence, confusing the reader and making errors more difficult to detect. This risk becomes more serious for retirement plans that not only contain extremely sensitive personal and financial data, but hold assets that participants rely on for their retirement security.
Errors may impact plan compliance, financial outcomes and liability. Plan document review is one example. AI may summarize plan documents quickly, but may also misinterpret provisions or rely on unchecked information. If an AI model misreads an adoption agreement or references an unchecked provision that appears in the agreement, its response could be wrong, which could lead to incorrect administration. While AI is a helpful tool, it should not be used without validation.
Reliable Data is Critical to Reliable Results
AI outputs are only as reliable as the information provided. If plan data, eligibility information or contribution formulas are incomplete or inaccurate, AI may build on those errors rather than identify them. As a plan sponsor, you should confirm that your plan data is complete, accurate and consistent with plan documents before relying on AI-generated results.
Fiduciary Responsibility Remains Unchanged
Using AI does not reduce your fiduciary responsibility or shift liability to the technology. Plan sponsors must remain responsible for plan operations, compliance with plan documents and adherence to applicable regulations.
While AI may support plan administration, outputs should be reviewed through appropriate checks, approval protocols and monitoring procedures. AI should not be used to make final compliance or fiduciary determinations without appropriate review against official plan documents, applicable provisions and relevant compliance standards.
Data Privacy and Cybersecurity Threats
Retirement plans contain extremely sensitive information such as Social Security numbers, compensation, financial information and addresses. Plan sponsors should understand what data is used in AI platforms, where it is stored, who has access to it and what safeguards are in place. Sensitive data should only be used in secure platforms with appropriate access controls, usage guidelines and vendor oversight.
As a plan sponsor, you should also review relevant Service Organization Control (SOC) reports and other cybersecurity documents from vendors to better understand how risks are being managed.
Vendor Oversight
A large portion of AI use in retirement plans occurs at the vendor level. Because of this, plan sponsors may not be able to control the use of AI directly. You should seek transparency from vendors regarding where and how AI is used.
Some important questions plan sponsors should ask vendors include:
- Where is AI being used within our plan?
- What participant data is being used?
- How is data protected?
- What review and testing processes are in place?
- How are outputs validated?
- Are any of these controls outlined in your SOC 1 report?
- How is the use of AI improving participant outcomes and plan administration?
- Are the AI tools internally developed, additional third-party tools or embedded within another platform?
Looking Ahead
AI will likely continue to play a larger role in retirement plan administration, but its use does not reduce a plan sponsor’s responsibility. You should understand how AI is being used, confirm that reliable data is being provided and maintain appropriate review procedures. Oversight, not automation alone, will determine whether AI supports effective plan administration.
We Can Help
Working with an experienced advisor can help plan sponsors identify gaps early, strengthen internal processes and improve coordination among internal teams and service providers.
Our team can assist you with audit readiness, internal control considerations, operational compliance and ongoing advisory support. Taking a proactive approach can help you better understand how AI is being used and whether appropriate safeguards are in place to support accurate and compliant plan administration.
Contact Us
If you have any questions, please contact your PKF O’Connor Davies client service team or:
Lauren Martin, CPA
Manager
lauren.martin@pkfod.com | 914.421.5621
Timothy J. Desmond, CPA
Partner
tdesmond@pkfod.com | 551.249.1728