I was struck by this article recently published in the WSJ on the use of AI for investor relations:

Investor relations departments at companies such as shoe brand Skechers USA and networking-systems and software provider Ciena have begun using generative artificial-intelligence to help prepare their earnings commentary.

Some have used generative AI to predict the questions analysts might ask, for example, and to ready the best answers….

Executives at many companies are using AI to refine word choice in their prepared remarks, for instance, in deciding whether to say the quarter was “strong” or “solid,” said Dan Sandberg, head of quantitative research and solutions at S&P Global Market Intelligence. The firm’s tool recently preferred “strong,” based on the earnings metrics of other companies that used the word on their earnings calls, he said. 

Generative AI can also read the harmonics in executives’ prepared statements on earnings, assessing them as upbeat, gloomy or something more measured…

As any securities litigator knows, these are exactly the kinds of nuances of communication that – when they become the subject of a securities fraud lawsuit – are routinely dismissed by courts as “puffery,” i.e., conveying no material information to investors. For example, a very quick Westlaw search yielded case after case treating the characterization of “strong” as inactionable. See, e.g., Ray v. StoneCo, 2024 WL 4308130 (S.D.N.Y. Sept. 25, 2024); Robeco Capital Growth Funds SICAV v. Peloton Interactive, 2024 WL 4362747 (S.D.N.Y. Sept. 30, 2024); Lloyd v. CVB Fin. Corp., 811 F.3d 1200 (9th Cir. 2016); Hawes v. Argo Blockchain, 2024 WL 4451967 (S.D.N.Y. Oct. 9, 2024).

Once again we see that securities cases are litigated in a world that is somewhat distinct from how corporations – and markets – actually function.