The First Amendment Does Not Protect Replicants

Article Source: Chapter in Social Media, Freedom of Speech, and the Future of our Democracy, Lee Bollinger and Geoffrey Stone, eds., Oxford University Press, 2022
Publication Date:
Time to Read: 2 minute read
Written By:



Often, the Constitution protects speech from censorship even if the speaker is not human. However, the Constitution may not bar all regulation of speech created by artificial intelligence-based systems (AI).


Policy Relevance:

AI-generated speech should not have full free speech rights.


Key Takeaways:
  • AI-based technologies could include a platform that automatically floods voters with AI-crafted tweets and videos; the content could be designed to produce victory for one candidate, based on data about what moves voters.
  • In several cases the Supreme Court has held that First Amendment rights of free speech protect speech regardless of the speaker’s identity; consistent with these cases, AI-generated content could not be regulated.
  • AI-generated political speech could harm democracy; neither we nor the framers of the Constitution understand the ramifications of unchecked machine speech.
  • "Replicants" select words amounting to meaningful speech, but this speech cannot reasonably be attributed to any human; that is, humans design systems for generating AI-created content, but do not control the systems' decisions about speech.
    • Similarly, gun manufacturers make weapons, but do not choose how the weapons are used.
    • An algorithm intended to create profitable categories for advertising created a category of "Jew Haters," which was quickly removed by Facebook executives.
  • Courts should follow settled constitutional doctrine, but should make exceptions when technology has changed the world fundamentally.
  • Democratic deliberation should be free of governmental control, preserving an equal right for every citizen to engage with every other citizen.
    • A content-generating AI is not a citizen entitled to equal standing with other citizens.
    • Content-generating AI is more like a foreign citizen or child, neither of which are fully protected by the First Amendment.
    • The free speech rights of listeners do not elevate all speech into protected speech.
  • AI-based content generators need not be banned, but are not entitled to full First Amendment protection.



Lawrence Lessig

About Lawrence Lessig

Lawrence Lessig is the Roy L. Furman Professor of Law and Leadership at Harvard Law School. His current areas of interest are Constitutional Law and Institutional Ethics, and he teaches and writes in the areas of constitutional law and the law of cyberspace. Professor Lessig has focused much of his career on law and technology, especially as it affects copyright. His current work addresses “institutional corruption”—relationships which, while legal, weaken public trust in an institution—especially as that affects democracy.