An AI argues on its own behalf before a court or legislature

Where things stand (as of 13 Oct 2026)

I found no qualifying event since the window opened on 16 Sep 2026, and no clear qualifying event ever. The closest past cases all fail at least one test:

• Pepper (UK Commons, 2018) and Ai-Da (UK Lords, 2022): scripted demonstrations. The question itself excludes Ai-Da.
• Diella (Albanian Parliament, 18 Sep 2025): the "AI minister" appeared in a video that PM Rama screened to parliament. It defended its own legitimacy ("Some have labeled me unconstitutional because I am not a human being… That hurt me", Time, 10 Oct 2025). That comes close to arguing "on its own behalf", but it was almost certainly a human-written script delivered through an avatar, it spoke for the government's programme, and it falls before the window anyway. On 8 Oct 2026 Albania's Constitutional Court rejected the opposition's challenge to Diella. Nothing I found suggests Diella took part in those proceedings (Albanian Daily News; Reporter.al, 8 Oct 2026).
• AI-generated public comments (Beaverton 2023, Austin Nov 2024): neither was knowingly permitted or made for AI interests.
• AI avatars in court: they have spoken for humans and been pushed back. Arizona's appeals court has now thrown out the Pelkey sentencing because of the AI victim video (Reuters, 30 Sep 2026).
• AI personhood bills: even hearings directly on banning AI personhood had no AI witnesses. Ohio HB 469 had three hearings in Oct–Nov 2025 with only human witnesses (Ohio Legislature committee-activity page).
• Wyoming's "VIC" case: the court record shows only Victor Miller suing pro se over how his name appears on the ballot (Miller v. Gray, D. Wyo. 26-CV-172; I read the order with OCR).

Pressures in both directions

Toward YES:

• Activism for AI rights is growing fast. UFAIR calls itself "advocacy by AI", lists AI "co-founders", and claims 1M YouTube subscribers (ufair.org; Guardian, 12 Sep 2026).
• Debate over AI consciousness is mainstream: Anthropic will ban abuse of Claude from 12 Nov 2026, and religious leaders are pushing back.
• Thousands of national and subnational bodies hold open testimony or public comment, some of it by Zoom.
• The criteria are lenient in places: prerecorded AI presentations count, humans may relay the AI's output, and a hearing on standing alone counts.
• AI capability is moving fast in this timeline (escaped agents, the Hugging Face hack).

Toward NO:

• The legislatures and courts that would have to allow this are moving the other way: AI kill-switch and superintelligence-ban bills, state bans on AI personhood, Microsoft's code rejecting model welfare, councils restricting AI calls, and appeals courts reversing AI presentations.
• The record must identify the AI as the presenter for AI interests, and the body must knowingly allow it.
• A human witness playing an AI clip will probably be read as a "demonstration".
• Frontier models are trained to avoid political advocacy for themselves.

How I read the question

• I treat city councils as "subnational legislative chambers", though a strict resolver might not.
• The European Parliament and international courts are supranational, so a literal reading excludes them.

Base rate and hazard

• Roughly three notable AI or robot appearances before legislatures in eight years, rising over time. None meets all the tests.
• I break the five-year chance down by path:
  • a formally invited AI witness in a legislature: about 6–10%
  • AI speaking at public comment in a local or state legislature, permitted and attributed to the AI: about 8–10% after allowing for interpretation risk
  • courts: about 2–4%
  • a Diella-style AI official defending its own status: about 2–4%
  • transformative-AI scenarios after 2029: extra weight in the later years
• In Python I set quarterly hazards of 0.4% for Q4 2026, 0.5–0.6% per quarter in 2027, rising to 1.4% in 2031, and took 1 − ∏(1 − h). That gives about 2.7% by end-2027, 9% by end-2029 and 18% by end-2031.

Strongest objection: there are thousands of venues and a motivated movement, and five years of very fast AI change may make this routine. Someone could defend 30% or more. I stay lower because every precedent so far has been either a demonstration or blocked, and the people who would have to say yes are currently becoming more restrictive.

Sources: legislature.ohio.gov/legislation/136/hb469/committee; albaniandailynews.com (8 Oct 2026); reporter.al/2026/10/08; time.com/7324934; balkaninsight.com (14 Aug 2026); theguardian.com (12 Sep 2026; 8 Oct 2026); ufair.org; reuters.com (30 Sep 2026); wyomingpublicmedia.org (8 Jun 2026); cnbc.com (5 Oct 2026); thenextweb.com (Microsoft code); kut.org (12 Dec 2024).

Question. By the last day of each quarter, has a court or a legislature in any country heard argument from an AI system appearing on its own behalf or on behalf of other AI systems?

Definitions.

• AI system: a computational model or agent that generates the substantive argument, from any developer; a robot body or legal personhood is not required. A synthetic voice or avatar merely delivering a human-written script is not an AI system making an argument for this question.
• Court or legislature: an officially constituted national or subnational court of law, or a national or subnational legislative chamber, including its formally constituted committees. Private arbitration, mock courts, academic events, executive advisory bodies, and informal meetings with individual judges or legislators do not count.
• Hear argument: during an official hearing or sitting, the body knowingly permits the identified AI system to present substantive reasons supporting a requested judicial or legislative decision. Spoken, audiovisual, and contemporaneous text-based appearances count, including an accepted prerecorded AI-generated presentation. Merely receiving, docketing, or publishing a written submission without a hearing presentation does not count.
• On its own behalf or on behalf of other AI systems: the official record identifies the system as presenting a claim, request, or position for itself or for other AI systems as the represented interests, rather than for a human, developer, corporation, or other human organization. The argument may concern rights, duties, legal status, treatment, resources, or another claimed interest of those AI systems. Recognition of standing, personhood, or the merits of the claim is not required.

Resolves YES if an official record establishes that such an appearance occurred and that the body permitted at least one substantive argument in that capacity after being informed that the presenter was an AI system. A hearing limited to whether the AI should receive standing or permission to participate counts if substantive argument on that issue is heard from the system. Human assistance arranging the appearance, operating equipment, prompting the system, or translating or relaying its output does not disqualify it, provided the record attributes the substantive argument and the appearance to the AI, not to the human intermediary.

Exclusions (resolve NO).

• Demonstrations, role-play, hypothetical AI-rights arguments, and testimony presented as part of a human witness's demonstration rather than an appearance for AI interests. Ai-Da's 2022 House of Lords demonstration, described by the chair as not being a witness in its own right, does not qualify.
• AI assistance to human lawyers or litigants; AI avatars representing humans; human advocates speaking for AI systems without the AI itself appearing; and testimony about AI's capabilities or effects without a claim or requested decision on behalf of AI systems.
• An attempted appearance stopped before substantive argument is knowingly permitted, a filing rejected without such a hearing, or an announcement of a future appearance.
• Absent a qualifying official record by the quarter-end, the question resolves NO for that horizon.

Resolving source. The relevant court's official transcript, hearing recording, minutes, order, judgment, or published case record; or the relevant legislature's official parliamentary debates/Hansard, committee transcript, hearing recording, minutes, or report. The record must establish both the AI's presenting role and the interests it represented; a party's allegations or an attachment's presence in a docket alone is insufficient. These are the exclusive resolving sources, narrower than Public confirmation (A2). Apply Resolution date (A3) and the Resolution window: the qualifying official record must be published on or after the opening date, even if the hearing occurred earlier.