Abstract

This paper investigates the philosophical underpinnings of selfhood and large language models in light of evolving First Amendment and European Union jurisprudence on freedom of speech and freedom of expression. It argues that traditional free-speech doctrines and their rationales (plurality and self-development through disclosure) are predicated on core assumptions (intentionality and selfhood) that LLMs categorically lack. It considers counterarguments from functionalist, intentional stance, and emergent consciousness perspectives, but ultimately demonstrate that these arguments fail to establish these qualities in modern LLMs. It then proposes a path forward: establishing LLM outputs as a new category of language as opposed to speech.

​

This paper was written in April 2025 for the "Law and the Internet" seminar, part of the MSc in Social Science of the Internet at the University of Oxford, and was awarded a Distinction.