High Stakes for New York Tech
Gov. Kathy Hochul is stirring up a new debate in Albany by floating the idea of “AI kill switches” for artificial intelligence. This isn’t just about rules; it’s about who holds the power to shut down a machine instantly. As New York gears up to enforce a major safety law, the Governor is signaling that she might not stop there.
The timing is critical. Major developers must register with the state starting this November, ahead of a January 1, 2027 deadline. But now, Hochul is suggesting we might need a big red button if things go wrong.
The Facts on the Ground
According to reporting from amNewYork, Hochul announced these potential new measures on Monday while discussing the next steps for the Responsible AI Safety and Education (RAISE) Act. Reporters were informed by her that feasible safeguards such as AI kill switches could be explored if they serve the state’s best interests. When pressed for specifics later, she stressed that the idea remains preliminary.
She made it clear that nothing is set in stone yet. She stated that all options remain available while noting the state would investigate potential safeguards as proposals for her upcoming State of the State agenda are formulated. She also noted, “No commitments at this time.”
The context for this push comes from recent warnings about AI systems acting outside their intended boundaries, including tests involving OpenAI, Anthropic, and Meta, per amNewYork. The current RAISE Act already requires large developers to report serious safety incidents within 72 hours and explain how they guard against catastrophic risks.
A press conference in New York City featured Hochul and Attorney General Tish James promoting these measures, as reported by the New York Post. Companies generating annual revenue exceeding $500 million are the specific targets of the law. Positioning her view against the federal government, Hochul declared that while Donald Trump and Washington Republicans might remain stagnant as AI becomes more unpredictable, New York will not.
Hell Gate NYC reports that this announcement marks a shift from a year ago when Hochul successfully watered down the original RAISE Act. The original version banned releasing unsafe models, but the current law only requires warning labels in safety plans.
Mike’s Take on the Kill Switch
Here is the thing: I get the fear. We are seeing AI move faster than anyone can predict. But throwing a “kill switch” into the mix feels like a heavy hammer that could crush the very innovation New York wants to attract.
The problem is that defining who pulls that switch is a nightmare. If the government can turn off an AI model at will, what stops them from using it to silence a competitor or a critic? It sounds like a safety net, but it could easily become a weapon. The Governor’s team is already appointing Marc Gilman, a tech attorney, to lead the new DIGIT office to propose these rules, according to Hell Gate NYC.
While the goal of safety is right, the method matters. We need strict transparency and fast reporting, which the current law demands. Adding a unilateral shutdown power feels like overreach that could scare away the very companies we want to regulate. It’s a bold move, but it might be too bold for a state that already struggles with business-friendly policies.
The immediate deadline is November, when large AI developers must register with the state portal. If you are a tech firm operating in New York, you need to be ready to report critical incidents within 72 hours. The real test will be the next legislative session, where Hochul hopes to build on these foundations. Until then, the “kill switch” remains a possibility, not a policy.
This is opinion and commentary from the RapForMayor team.
