Bloomberg Law examines how artificial intelligence is transforming ERISA litigation on both the plaintiff and defense sides of the bar — and speaks with Darrow's Michael Harary and Harel Fisher to understand what AI-driven legal research now makes possible at scale.
Plaintiff attorneys are using AI to scan the universe of ERISA-covered plans, identifying outliers and anomalies that may indicate mismanagement of retirement assets or improper handling of forfeiture funds. A key data source is the Form 5500 — the standardized annual filing all large ERISA-covered plans must submit — which gives AI tools a structured dataset to analyze at volume. Darrow's Michael Harary, director of litigation partnerships, describes what this looks like in practice: the platform analyzes more than 200,000 plan sponsors, 60,000 retirement funds, and over $6 trillion in plan assets, applying benchmarks to each fund and flagging deviations that may warrant legal action. Work that once required weeks of manual effort now happens systematically and continuously.
Defense attorneys are finding parallel value. Rather than manually reviewing financial data to build a benchmarking defense for a motion to dismiss, AI can quickly assemble comparisons of fees and performance across investment options — giving defense counsel a data-backed foundation in a fraction of the time. AI is also being used across both sides to assist with drafting court filings, though the article notes a growing number of incidents involving AI-generated hallucinations that have led to sanctions.
Harel Fisher, who leads Darrow's ERISA Research Lab, points to ERISA-covered health plans as the next major frontier for litigation — a conclusion drawn from early court rulings and data trends Darrow is already tracking. For plan sponsors and their counsel, that kind of forward-looking signal is increasingly valuable: getting ahead of emerging exposure is far preferable to responding to it after a complaint is filed.
The piece also highlights more tactical applications becoming standard in the field. AI can generate copycat complaints by identifying recent class actions that survived motions to dismiss, and can be used to prepare for depositions by surfacing questions attorneys might not have thought to ask. Adam Garner of The Garner Firm describes AI as a useful supplemental tool, while cautioning that over-reliance without verification is a serious professional risk.
The structural risk that receives the most attention is hindsight liability. Rick Nowak of Mayer Brown raises the concern that as AI makes it faster and easier to reconstruct what a fiduciary could have done differently, it may enable litigation that second-guesses reasonable decisions made in good faith — precisely the kind of outcome ERISA's framework was not designed to incentivize. The Department of Labor appears to be watching: the article notes a recent proposal to establish a safe harbor for fiduciaries who include alternative assets in 401(k) plans, signaling an effort to provide greater clarity as the litigation environment intensifies.
