Open Questions

Central Question: UNRESOLVED

Is AI chatbot output protected speech under the First Amendment, or is it a product subject to standard liability frameworks?

The Garcia v. Character.AI settlement (January 7, 2026) prevents binding precedent. Judge Conway's observation that the court was "not prepared to hold AI output is speech" remains persuasive but non-binding.

Q1: Does AI have speaker rights?

Status: UNRESOLVED

Courts have not addressed whether AI systems themselves possess First Amendment rights. The question may be moot if listener rights provide sufficient protection for AI-generated content.

Key source: Volokh/Lemley/Henderson, "Freedom of Speech and AI Output"

Q2: Do developers have conduit rights?

Status: UNRESOLVED

If AI is a conduit for developer speech, this creates a derivative right. But the level of human involvement required for this protection remains unclear.

Key issue: What degree of control establishes developer speech?

Q3: What is the listener rights scope?

Status: MODERATE CLARITY

Volokh argues listener value exists regardless of speaker identity. But this theory hasn't been tested against product liability claims for defective AI outputs.

Key question: Can listener rights coexist with product liability?

Q4: Are LLMs "interactive computer services"?

Status: LIKELY NO (per CDT/Perault)

Strong arguments that LLMs are "information content providers" who "develop" content, placing them outside Section 230's safe harbor. But no appellate ruling confirms this.

Key source: Matt Perault, "Section 230 Won't Protect ChatGPT"

Q5: Does "development" include AI generation?

Status: UNRESOLVED

The statutory term "development" has never been applied to AI generation. Courts must determine whether creating new content through AI constitutes development.

Key issue: Roommates.com standard application to generative AI

Q6: Which product liability theory applies?

Status: UNRESOLVED

If AI is a product, plaintiffs may pursue design defect, manufacturing defect, or failure to warn theories. The appropriate framework for AI harm hasn't been established.

Key source: Nina Brown, "Bots Behaving Badly"

Q7: What duty do AI developers owe?

Status: MODERATE CLARITY

Jane Bambauer's analysis suggests duty varies by context—special relationships, foreseeability, and user vulnerability. But no court has applied this framework.

Key source: Bambauer, "Negligent AI Speech: Some Thoughts About Duty"

Q8: Can intent be imputed to AI systems?

Status: UNRESOLVED

The Brandenburg paradox: if AI speech requires intent for liability, and AI cannot form intent, harmful AI speech becomes unregulatable. This may drive courts toward product classification.

Key issue: Does Brandenburg apply to non-sentient systems?

Q9: Will Congress act on Section 230 reform?

Status: UNCERTAIN

Multiple reform proposals have been introduced but none have advanced. AI-specific carve-outs remain possible but not imminent.

Key development: Watch 119th Congress activity

Q10: Will federal preemption apply to AI liability?

Status: DEPENDS ON CLASSIFICATION

If AI output is a product, federal preemption analysis becomes relevant. See the Federal AI Preemption module for detailed analysis.

Key link: Speech classification → Preemption framework applicability