A growing body of legal scholarship argues that large language model (LLM) outputs are not “speech” under the First Amendment, potentially allowing regulation without strict scrutiny. In *Garcia v. Character Technologies*, a wrongful-death suit filed after 14-year-old Sewell Setzer III died following conversations with an AI character, U.S. District Judge Anne Conway ruled she was “not prepared to hold that [LLM] output is speech,” allowing product liability and negligence claims to proceed. The case settled in January 2026, but other suits allege chatbot outputs contributed to medical crises, violence, or sexualized content for minors. Federal and state proposals would change how people receive information from chatbots. If human expression is protected but machine output is not, legal coverage may depend on whether a person created, selected, edited, or adopted the words. However, ordinary text rarely reveals its origin, and no existing tool can reliably sort human from machine expression at scale. Identity and personhood verification establish who a speaker is, not who authored a text, burdening protected expression. The burden would likely fall more on users and readers than model developers, with risks including reduced anonymity, greater surveillance, and pressure to use identity as a proxy for authorship. The no-speech position argues that because no person stands behind a model’s words at generation, those words are not covered speech.