Investment Firms Prioritize AI Risk Warnings Over Usage Claims

The analysis, published in Frontiers in Artificial Intelligence, reviewed Form ADV Part 2A brochures from 388 SEC-registered firms. While 22.7% of the weighted sample reported AI usage, 27.6% included explicit risk language. This discrepancy suggests that firms are more inclined to hedge against potential technological fallout than to detail specific operational or investment applications. Larger firms demonstrated a higher propensity for disclosure, with 43.8% of the highest asset-management group mentioning AI, compared to only 10.4% among smaller entities.

Identifying these disclosures remains a complex task for automated systems. Researchers utilized GPT-4o to categorize documents, discovering that narrow keyword searches often miss nuanced descriptions of machine learning. When comparing brochures to website content, the study found that disclosure rates appear similar once document length is accounted for, though website classification remains less reliable. Despite concerns regarding potential "AI-washing," the researchers found no immediate evidence of unsubstantiated promotional claims mirroring recent SEC enforcement actions against firms like Delphia and Global Predictions. The study highlights that silence in a filing does not equate to an absence of technology, as many firms may be experimenting with AI behind the scenes without formalizing these activities in regulatory disclosures.