The advent of generative AI (GenAI), which can produce billions of words daily, is forcing a profound reexamination of the First Amendment’s protection of “freedom of speech.” Written in 1791 to safeguard human expression, this constitutional right is now being claimed by AI companies seeking immunity from liability for the harmful outputs of their products.

The central question is whether the algorithmic output of a large language model (LLM) – a statistical calculation rather than a human expression of belief or intent – qualifies as “speech” under the First Amendment. Examining the traditional justifications for protected speech – self-expression and the “marketplace of ideas” – the chapter argues that neither rationale neatly applies to content produced by nonsentient machines. It then explores two proposed legal standards: the “speech certainty” standard, which grants protection only when the content is predictable and controllable by a human author, and a human-centric approach requiring intentionality, self-awareness, and human expression. Granting full First Amendment protection could immunize AI companies from liability for a range of harms.