OpenAI has sent a letter to the New York City Council recommending safeguards against AI risk, according to The Information. The company is taking its policy pitch past Washington and the statehouses and straight to a city council. That’s unusual, and it tells you where the AI rulemaking fight is heading.
The Information’s report gives only the basics: OpenAI wrote to the council, and the letter proposes safeguards. The full text of the recommendations hasn’t been made public. So this briefing covers what we know, why it matters and what to watch next.
Situation Assessment
What this is: the leading frontier AI lab is lobbying at the city level.
That’s a big change in approach. Until now, AI labs have spent most of their policy effort in three places:
- Washington, pushing for one light-touch national framework
- State capitals, mostly California and New York, where the frontier model safety bills have been written
- Brussels, where the EU AI Act is now being rolled out in phases
Cities have mostly been left out of that conversation. A letter to the NYC Council means OpenAI now sees local government as a place where it has to show up.
Why New York City Matters
New York isn’t a random choice. The city already has one of the first AI-specific rules in the U.S.: Local Law 144, which requires bias audits for automated hiring tools. It took effect in 2023, and it showed that cities can and will regulate AI directly.
The city also sits inside a state that has been aggressive on AI safety. New York’s RAISE Act targets developers of the most powerful models. That puts the city’s lawmakers in a crowded, active policy environment.
New York is also a huge market. It’s full of banks, media companies, hospitals and city agencies that are adopting AI tools. Rules written there tend to spread, either because other cities copy them or because companies adopt them nationwide so they don’t have to run a separate version for one city.
Reading the Move
OpenAI has a clear track record here. When California debated SB 1047 in 2024, the company opposed it. In 2025 it wrote to Governor Newsom asking that companies already following federal or international safety frameworks be treated as compliant with state rules. The pattern is the same each time: OpenAI supports safeguards in principle and tries to shape how they’re built.
I’d read the NYC letter the same way. When a company offers “recommended safeguards” before a legislature writes its own, it’s doing two things:
- Setting the template. The first detailed proposal on the table often becomes the draft everyone edits.
- Heading off a patchwork. AI companies’ biggest fear isn’t regulation itself. It’s dozens of conflicting local rules. Getting in early lets OpenAI push for consistency.
None of this is necessarily cynical. Lawmakers often lack technical expertise, and input from the people building the systems has real value. But readers should keep in mind that the advice comes from a company with a stake in the outcome.
Implications for Practitioners
If you build or deploy AI in New York, or anywhere else, watch these points:
- Local compliance is real now. Don’t assume AI rules only come from federal or state law. Your city may add its own requirements.
- Audit and disclosure rules are spreading. NYC’s hiring-tool law set a precedent. Expect similar requirements in other high-stakes areas.
- Preemption is the wild card. Washington has been pushing to override state and local AI laws. How that fight ends decides whether city rules like these actually stick.
- Document your safeguards now. Whatever framework wins, teams with written risk assessments and testing records will adapt faster.
What Comes Next
The key open question is what’s actually in the letter. Whether OpenAI pushed for tougher rules or softer ones, and for which uses, will show how much of this is safety advocacy and how much is positioning. Watch whether the council introduces legislation that echoes the letter’s language. And watch whether other labs, such as Anthropic and Google, start sending their own letters to city halls.
The AI policy map just got more local. More details are available in the original report from The Information.