Author: Nita Farahany Source: https://nitafarahany.substack.com/p/the-statute-that-couldnt-stretch Published: March 4, 2026
A class essay by Nita Farahany, published March 4, 2026, as part of her Advanced Topics in AI Law and Policy course (Class 7.2 of roughly 30). The essay presents the Telephone Consumer Protection Act (TCPA) and the Supreme Court's 2021 decision in Facebook v. Duguid as a case study in how a statute written for an earlier technology can be narrowed by interpretation so that modern AI-driven communication mostly falls outside it. Farahany frames the example as a lesson for any attempt to use existing statutes against new persuasion technologies, and uses it to set up S. 1629 as a drafting response.
Summary of argument
Farahany describes the TCPA as the most plaintiff-friendly consumer-protection statute in the country until Facebook v. Duguid (2021) narrowed its scope. The argument she draws is that a statute written for one technology can be made functionally irrelevant by reading its definition narrowly enough to track only the original technology. She characterizes this not as bad faith but as textualism applied to a roughly 30-year-old definition, and concludes that any statute attempting to reach AI persuasion must be drafted to anticipate technological change or it will be narrowed out of relevance the same way.
Key claims
The TCPA, enacted in 1991, prohibited "automatic telephone dialing systems" (ATDS), which it defined as devices that "store or produce telephone numbers to be called, using a random or sequential number generator." It set statutory damages of $500 per violation, rising to $1,500 if willful. By 2018, TCPA suits were the leading federal consumer class-action category.
In Facebook v. Duguid (2021), decided 9-0 with an opinion by Justice Sotomayor, the Court narrowed "ATDS" to require random or sequential number generation. Under that reading, modern systems that dial from stored lists of customer numbers are not ATDS, with the result that the majority of modern automated calls and texts fall outside the statute. Farahany argues this matters for AI persuasion because personalized AI persuasion systems virtually never use random number generators; they target known individuals with personalized messages. On her account, Duguid removed the most powerful federal statute from the field at the moment the field most needed it.
Farahany presents S. 1629, the Stopping Loot Boxes Act (2025), a Markey/Hawley bill responding to the FTC loot-box record, as a legislative response drafted to avoid the Duguid problem. She notes that the bill is explicitly defined to reach personalization architectures rather than only "automatic dialers," using forward-looking definitional language ("any system that adapts content presentation based on user behavior signals"). Whether that language survives a Duguid-style narrowing is, in her framing, the open bet. Generalizing the lesson to the rest of Week 7, she argues that a new AI-persuasion statute must answer the Duguid problem at the drafting stage: the fix is not to enumerate technologies but to define behaviors and harms, and to build in interpretive instructions that resist technology-specific narrowing.
Relationships
- part-of: Nita Farahany Advanced Topics course (Class 7.2 of ~30)
- related: Regulating Under Uncertainty
- previous: Inside My Advanced Topics Class 7: The Persuasion Exchange (Farahany, March 2026) next: Inside My Advanced Topics Class 7.3: Aristotle's Algorithm (Farahany, March 2026)