New York governor says she's using AI to analyze ‘every single rule' in the state

New York Governor Kathy Hochul is taking a pragmatic approach to artificial intelligence that might surprise those who just read about her administration's recent pause on new AI data centers. While she is cautious about the infrastructure boom, she is actively deploying these same tools to modernize the state's legal framework, according to reporting by The Verge.

During an appearance on Bloomberg's Odd Lots podcast, Hochul revealed that her team is using AI to analyze every single rule, regulation, and policy in New York. The goal is straightforward but ambitious: identify and flag outdated legislation that no longer serves the public interest or makes logical sense in a modern context.

The results of this digital audit have been both impressive and somewhat embarrassing for the state. Hochul highlighted specific examples of antiquated laws that the AI flagged, including a twenty-five dollar fee required to take a dog hunting. It sounds like a relic from a different era, but it is still technically on the books.

Even more striking is the discovery that pregnant people currently need a permit to work after midnight. This regulation is a stark reminder of how slowly legal codes evolve compared to social norms. The AI did not just find typos; it found structural absurdities that human reviewers might have overlooked due to sheer volume.

Hochul noted that conducting this review manually would have taken five years at the staff level. This efficiency gain is the real story here. It demonstrates how generative AI can handle bulk processing of unstructured text, a task that is traditionally slow, expensive, and prone to human error.

However, this approach also highlights a critical limitation of current AI models. These systems are excellent at pattern recognition and keyword matching, but they lack true contextual understanding. A law requiring a permit for pregnant workers might have been intended to protect health in a pre-industrial setting. Removing it requires human judgment about intent, not just syntax.

The broader implication is that government agencies will increasingly rely on AI for compliance and regulatory cleanup. This could lead to faster updates to laws but also risks removing protections if the AI misinterprets the original purpose of a regulation. We need to see how states balance speed with careful legal review.

What this means for you: If you work in compliance, legal tech, or policy, start experimenting with AI to audit your own internal documentation. You can use large language models to find contradictions or outdated clauses in your company's handbooks. Try this prompt with your AI assistant to get started: "Review the attached policy document and list any clauses that reference technologies, laws, or practices that seem outdated or contradictory to modern standards. Explain why each item might be obsolete."