What AI Analysis Can Reveal About Securities Class Actions
Law360, 2026
Securities class action litigation depends on complex assessments, including the evaluation of comparable cases and the prediction of settlement outcomes. Historically, these analyses have relied on structured data and limited manual review of complaint text, constraining the range of observable case characteristics and the insights that can be drawn from unstructured legal filings.
In an article published in Law360, Analysis Group Managing Principal Mark Howrey and Vice President Emma Xiaoxiao Dong examine how AI-assisted review of complaint text can enhance securities litigation analysis. Focusing on two key areas – identifying comparable cases and predicting settlement outcomes – the authors show how AI can expand the scope and depth of these analyses.
Drawing on a database of more than 700 federal securities class action settlements, the authors demonstrate how AI can systematically extract case characteristics – such as the nature of allegations and corrective disclosures – that enable more robust identification of comparable cases. They find that differences in the type and substance of corrective disclosures are associated with variation in settlement outcomes – for example, disclosures tied to enforcement actions are generally linked to higher settlement percentages than those involving investigations alone.
The authors also find that sentiment-based measures of complaint language provide additional explanatory power in modeling settlement amounts, reducing prediction errors beyond models based solely on damages. At the same time, they emphasize that the complexity of securities litigation cannot be fully captured through AI-assisted analysis alone, as certain key drivers of settlement outcomes – such as defendants’ insurance structures – are not observable in public data. Accordingly, AI-based insights are best viewed as a complement to, rather than a substitute for, legal and economic judgment.