
Tomorrow, the New York City Council holds a hearing. On Monday, October 5, 2026, the Council is demanding that AI companies testify under oath.
The Council will review a set of AI regulation bills there. One of them would require AI models marketed, offered for sale, sold or deployed in the city to have a way for a human to shut them down. If you hold or watch AI stocks, this is worth knowing.
◆ The Hearing at a Glance
When: Monday, October 5, 2026
Format: Committee of the Whole, with all 51 Council members taking part
Bills: an initial list of 10, released September 25, with more to come
Invited: the CEOs of Anthropic, OpenAI, Google, SpaceXAI and Meta
Testimony: under oath, in public (SpaceXAI's attendance isn't confirmed)
1. A Hearing With the Whole Council
This isn't an ordinary committee hearing. It's a Committee of the Whole, so all 51 Council members take part and weigh the risks AI poses to New Yorkers.
The Council itself calls it a rare event.
Council Speaker Julie Menin released an initial list of 10 AI bills on September 25. Her message, in essence: being pro-innovation and pro-safety can go together, and should.
If New York wants to remain the capital of AI, responsibility has to come with it. The Council also said more bills are on the way, including one on deepfakes.
2. The Four Bills That Matter Most
Of the 10, four stand out for investors. All of them are still proposals. None has been passed. What follows is what would apply if they pass.
◆ Four Proposed Bills
Third-party validation
To market, offer for sale, sell or deploy an AI model in the city, it needs outside validation. Without validation or a shut-down capability: $25,000 per violation.
Kill switch
Humans must be able to stop the AI. This sits in the same bill as validation.
Whistleblower rewards
Report a violation of the bill's AI provisions and receive 25% of what is recovered, or 50% if the city formally authorizes you to pursue the case. The Council calls it a first-in-the-nation approach.
Private right of action
Harm in the city that was foreseeable, with no reasonable safeguards in place, can be taken to civil court. A separate bill.
First is third-party validation. To market, offer for sale, sell or deploy an AI model in New York City, an outside validator must first check data quality, bias, outputs, privacy and security.
An AI model marketed, offered for sale, sold or deployed without validation, or without a shut-down capability, draws a $25,000 penalty per violation. A validator who falsifies a validation faces the same amount.
Second is the kill switch. AI deployed in the city must let a human stop it, and the validator checks that this capability exists. It's part of the same bill as the first.
Third is whistleblower rewards. Report a violation of the bill's AI provisions and, once money is recovered, the reporter receives 25% of it. If the city designates the reporter (or the reporter's attorney, if the reporter isn't one) to bring a civil action, or authorizes the reporter to serve a notice of violation, the share can reach 50%. The Council calls this a first for the nation.
Fourth is the private right of action. A company that offers an AI model for commercial or public use may fail to put reasonable safeguards in place. A third party may then exploit that gap, or bypass the safeguards ("jailbreaking"), and foreseeable harm may follow in the city. The victim could then sue in civil court.
The package also includes a bill on city contracts. A contractor or city agency that learns of a reportable AI safety incident would have 24 hours to notify the city's Cyber Command.
◆ From Bills to Hearing
Sept 25 Speaker Menin releases an initial list of 10 AI bills and announces more to come.
Sept 28 The Council issues a subpoena to SpaceXAI. Meta, OpenAI, Google and Anthropic had already agreed to attend.
Oct 5 The Committee of the Whole hearing takes place.
3. Why Subpoenas Were Needed
The Council asked the CEOs of five companies to attend: Dario Amodei of Anthropic, Sam Altman of OpenAI, Sundar Pichai of Google, Elon Musk of SpaceXAI and Mark Zuckerberg of Meta. At first, reports said most wouldn't come.
So the Council reached for the subpoena.
Meta was first to agree, confirming that a member of its senior leadership would appear. OpenAI, Google and Anthropic agreed only after the Council threatened subpoenas. Anthropic reportedly confirmed just hours before its subpoena was due.
The Council issued a subpoena to SpaceXAI on September 28, after the company gave no answer. If it doesn't comply, the Council can ask the New York State Supreme Court to enforce it.
Anthropic, OpenAI, Google and Meta are confirmed to testify publicly under oath. Reports differ on who will actually appear, so a CEO in person can't be assumed. SpaceXAI holds a subpoena, and as of October 4 its attendance is unconfirmed.
◆ Three Layers of Government
Federal. In July 2025 the Senate voted 99-1 to strip out a 10-year limit on state AI rules tied to BEAD broadband funds. In mid-September, President Trump called AI safety threats a "hoax."
New York State. The RAISE Act: the state said it would begin directing large developers to register in November 2026, and the law takes effect January 1, 2027.
New York City. An initial list of 10 bills, with the hearing on October 5.
4. Why the City Council Is Acting
Start with Washington. In July 2025, the Senate voted 99-1 to remove a provision that would have limited states' AI rule enforcement for 10 years, a limit tied to federal BEAD broadband funds.
New York State has already passed the RAISE Act. It covers developers of frontier models trained with more than 10^26 operations that are developed, deployed or operated in New York, in whole or in part. Large developers, with more than $500 million in prior-year revenue counting affiliates, face additional duties.
The state said it would begin directing large developers to register in November 2026.
The law's main compliance date is January 1, 2027. From then, frontier developers covered by the act, regardless of revenue, must report a critical safety incident within 72 hours of determining that one occurred, or of learning facts that reasonably indicate it did.
If an incident poses an imminent risk of death or serious injury, the developer must disclose it within 24 hours to an appropriate authority, which may include a law enforcement or public safety agency with jurisdiction. The city's bills are separate, and each has its own scope: sales and deployment of AI, harm from third-party misuse, and safety incidents under city contracts.
Meanwhile, in mid-September, President Trump called AI safety threats a "hoax" and signaled that no extra regulation is needed.
◆ Market Numbers (as of September 14)
about 60%
Philadelphia Semiconductor Index, year-to-date gain
$795B → $1.08T
Capital spending by five big tech companies, 2026 to 2027 (BofA estimate)
5. Check the Market Before the Bills
Before the bills, look at the market. As of Monday, September 14, the Philadelphia Semiconductor Index was up about 60% for 2026. The S&P 500 and the Nasdaq Composite both sat below their record highs that day.
The weekend before, Amodei and other AI figures had spoken in favor of a slower pace of development. On Monday, chip and AI-related stocks swung hard.
Capital spending by Meta, Microsoft, Alphabet, Amazon and Oracle is estimated by BofA Global Research at about $795 billion in 2026 and about $1.08 trillion in 2027.
Any talk of AI slowing down puts that spending plan in question, which is why the market reacts so sharply.
6. How Two Investment Professionals See Regulation
Two voices from that week's coverage. Erik Kratz, chief investment officer at Arena Private Wealth, said in essence that a credible safety framework can make long-term capital spending easier to justify.
He also explained that chip stocks take the hardest hit from such debates because they're the layer most exposed to a slowdown in progress.
Michael Bruun of Goldman Sachs Alternatives said investors must weigh whether regulation gets stronger or weaker and whether geopolitical tension grows. Even so, he said he still sees chances to put capital to work at the epicenter of AI.
These are the views of two people, not a market consensus. There isn't enough evidence to say regulation won't break the AI investment cycle.
◆ Three Things to Watch
1 Which companies send whom, and what they say
2 Whether SpaceXAI complies with the subpoena
3 Whether core provisions such as the kill switch and third-party validation survive intact or get changed
7. What to Watch From Tomorrow
First, a hearing is a place to listen. No bill passes that day, and no vote date has been confirmed.
Whether the bills pass as written is also undecided, so the outcome can't be called in advance.
There are three things to watch: first, which companies send whom and what they say; second, whether SpaceXAI complies with the subpoena; and third, whether core provisions such as the kill switch and third-party validation survive intact or get changed.
If rules at the city level spread to other cities and states, compliance costs for companies could rise. That's why this is worth following.
This isn't investment advice. You know that, right? Let's each protect our own hard-earned money.
That's all for this post. See you next time.
Related posts
- Jobs Shock, Yet the Nasdaq Hits a Record? AI Bubble, Rates, and Oil in One Read
- Expert-Recommended AI Semiconductor Stocks for October 2026: NVIDIA, Broadcom, TSMC. Should You Buy Now?
- OpenAI Funding Stalls, AI Stocks Sink: If Money Dries Up?
🌐 Korean version: 소환장까지 날아간 뉴욕시 AI 청문회, 규제가 AI 주식을 흔들까?
🌐 Japanese version: 召喚状まで飛んだNY市のAI公聴会、規制はAI株を揺らすか?
No comments:
No comments yet. Want to leave the first one?
Post a Comment