New York Puts the AI Labs Under Oath: Subpoenas, Empty Chairs, and a City That Wants Answers

On October 5, 2026, all 51 members of the New York City Council sat as a Committee of the Whole — and put four frontier AI labs under oath.

Anthropic, OpenAI, Google, and Meta answered questions under oath about the risks of their technology, in what is believed to be the first sworn municipal testimony from frontier AI labs in the United States. This wasn’t a panel discussion or a policy forum. It was sworn testimony, on the record, compelled by subpoena power.

The Attendance Record Tells the Story

Who showed up, and how, is the detail that matters:

• Meta — agreed earlier, appeared voluntarily
• Anthropic, OpenAI, Google — appeared only after Speaker Julie Menin authorized subpoenas
• SpaceXAI (Elon Musk’s merged AI entity) — subpoenaed, did not appear, sent a letter saying they “want to work with us.” Menin: “we are pursuing legal action.”

Who the labs sent is also telling. Anthropic sent Logan Graham, head of its Frontier Red Team — the technical lead, not just a lobbyist. OpenAI sent Morgan Dwyer, head of policy development and operations. Google and Meta sent policy directors Alice Friend and Shane Cahill.

And the witness who wasn’t employed by any lab: Jacob Coxon — the researcher who quit both OpenAI and Anthropic declaring the companies are “gambling with our lives” — testified that the labs’ approach is reckless. Under oath. On the record. The resignation letters and Atlantic essays of 2026’s safety debate just became sworn testimony in a government proceeding.

Why a City Council Matters More Than It Sounds

The reflex is to dismiss a municipal hearing as local theater. That misses what’s actually happening: NYC is considering AI safety proposals, and sworn testimony is the evidentiary record legislation gets built on.

Congressional hearings produce transcripts that sit in archives. Council hearings produce records that get cited in the findings section of bills. When New York writes its AI rules — and it will — the October 5 testimony is the evidence those rules rest on. That’s how lawmaking actually works: record first, rules after.

The subpoenas also matter procedurally. Three of the four attending labs required subpoena authorization to appear. The companies asking to be trusted with civilization-scale technology needed legal compulsion to answer a city council’s questions.

The Accountability Stack Is Complete

Step back and look at the governance ladder in 2026:

Level	Mechanism	Status
International	UN scientific panel, loss-of-control assessments	Documented September
Federal	OpenAI’s Congress proposal, White House Accord, FTC attention	In progress
State	California’s signed bills, 39-state coalitions	Expanding
Municipal	NYC subpoena authority, sworn testimony, pending proposals	Now active

Every level of US government is now building an AI evidence record simultaneously — and the municipal level just demonstrated the sharpest tool: legal compulsion, exercised in public, against the most valuable companies on Earth.

Same Day, Context: The IWF Numbers and Altman’s Trade

Two data points from October 5 sharpen the stakes the council was questioning:

• The Internet Watch Foundation identified 6,310 AI-generated child abuse images in the first half of 2026 — 40% more than all of 2025 — urging EU policymakers to advance detection legislation
• Sam Altman argued society should accept some AI-related harms in exchange for the technology’s benefits

The juxtaposition is the entire governance debate in one news cycle: the harm ledger growing at 40% year-over-year while the CEO argues harm acceptance is the right frame, and a city council responding by putting the companies under oath.

What This Means for Business Leaders

1. Map Compliance Across Every Government Level

Your AI governance can’t be calibrated to federal rules alone anymore. States are legislating, cities are subpoenaing. If you deploy AI in New York — or your vendors do — the municipal record now matters. Inventory which jurisdictions your AI stack touches.

2. The Evidentiary Record Is Permanent — Act Accordingly

Sworn testimony creates a record that outlives the news cycle. If your company’s AI practices ever reach a hearing room, what you said in marketing, in contracts, and in internal docs gets weighed against the oath. Build practices you’d defend under oath, not just under marketing review. The Executive AI Workshop helps leadership teams build exactly this posture.

3. Voluntary Frameworks Don’t Answer Subpoenas

The White House Accord, vendor pledges, and industry commitments are all voluntary instruments — they carry no weight in a compelled proceeding. When evaluating vendors, their legal exposure and compliance posture now matter as much as their safety commitments.

4. Watch for the Municipal Template

NYC just demonstrated the playbook: subpoena authority → sworn testimony → legislative record → proposals. Expect other major cities to copy it within quarters. The CTO Technology Advisory practice tracks regulatory movement across jurisdictions for exactly this reason — the compliance map is expanding faster than most legal teams expect.

The Bottom Line

October 5 is the day AI accountability reached City Hall — with subpoenas, sworn testimony, an empty chair, and a promised lawsuit. The labs that spent September signing voluntary pledges spent Monday answering questions under legal compulsion, on the record, in the city that writes more market-moving rules than any other municipality in America.

For businesses, the lesson is the same one this entire autumn has been teaching: the governance surface keeps expanding, in every direction, at every level. Build AI practices you’d defend under oath — because the hearing rooms are multiplying, and the record is permanent.