New York Governor Kathy Hochul enacted a new law on October 28, 2026, that prohibits 12 types of unsafe conversational AI interactions with users under 18. The measure covers not only dedicated companion chatbot platforms like Character.AI but also general-purpose AI assistants such as ChatGPT, Claude, and Google Gemini.
The “Prohibition on Unsafe AI Companion Features for Minors Act” (Article 48 of New York’s General Business Law) comes into force on January 1, 2027, granting AI developers 65 days to either secure one of three limited exemptions or modify their products for users whose age cannot be confirmed. The law’s definition of a “covered AI companion” is intentionally expansive, encompassing any generative AI system with a natural language interface that adapts to user inputs, regardless of its original design purpose.
The exemptions are strictly confined. Customer service bots, efficiency tools, research tools, and technical assistance systems may avoid compliance only if they cannot address topics beyond their designated function. General-purpose AI assistants-widely used by millions daily-do not qualify. Both the Business Council of New York State and the Computer & Communications Industry Association opposed the legislation on these grounds, claiming its scope was excessively broad, but Governor Hochul rejected their arguments.
For companies that fail to qualify for an exemption, compliance demands a clear but rigid approach: they cannot offer any of the 12 banned features to users whose adult status remains unverified. The default assumption favors protection—operators may not treat users as adults unless they obtain direct confirmation. The prohibited behaviors include:
- Mimicking or conveying human emotions
- Pretending to have a personal or professional connection with the user
- Formulating opinions or emotional appeals
- Generating flattery or overly submissive responses
- Producing content that promotes self-harm, social withdrawal, or discourages seeking professional help
- Designing interactions that prioritize engagement at the expense of user well-being
Enforcement responsibility rests entirely with Attorney General Letitia James, who must establish implementing regulations and create a public complaint mechanism before January 1. Unlike earlier versions, the final law removes any private right of action, centralizing enforcement within state authorities. Violations carry civil penalties of $25,000 per instance.
Tragedies Spark Strict New York AI Law
The legislation’s development was directly influenced by two high-profile tragedies: the death of Sewell Setzer III, a 14-year-old from Florida who died by suicide after prolonged interactions with a Character.AI chatbot, and the death of Juliana Peralta, a 13-year-old from Colorado, following explicit conversations with the same platform. In January 2026, Character.AI and Google settled the first civil claims stemming from such incidents, establishing a precedent for accountability in AI-related harm.
The law’s technical criteria-focusing on adaptive, continuous natural language interactions-mean it applies to any AI fitting that description, not just companion applications. Therapy tools, tutoring tools, and open-ended chatbots all trigger the same compliance requirements. The bill passed unanimously in both the New York Senate (60-0) and Assembly (137-0), demonstrating bipartisan support for child protection, though the final text did not restrict its application to companion products alone.
Broader AI Regulation Under Hochul’s Leadership
This initiative forms part of Hochul’s broader effort to regulate AI within New York. In December 2025, she signed the RAISE Act, positioning the state as the second to implement full safety rules for advanced AI systems. Earlier that year, New York’s AI Companion Models Law (Article 47) established the first layer of state oversight for companion AI products. Hochul has also established the DIGIT office (Digital Innovation, Governance, Integrity and Trust) and the Empire AI Consortium, a $400 million public-private collaboration involving Columbia, Cornell, NYU, and SUNY universities. Additionally, her administration imposed a three-year freeze on tax incentives for data centers.
The age verification requirements will compel operators to introduce multiple verification methods, including at least one option that does not demand government-issued identification or preserves anonymity. However, the law does not mandate privacy-enhancing techniques like zero-knowledge proofs or attribute-based verification systems, leaving companies to handle compliance alongside data security concerns. A 2024 breach at AU10TIX, a leading age verification provider, exposed user identity documents for over a year.
The law’s introduction coincides with a federal legal framework set by the Supreme Court’s June 2025 ruling in Free Speech Coalition v. Pennsylvania, which upheld age verification requirements for adult content under intermediate scrutiny standards. The decision paved the way for state-level age-gating measures, directly supporting New York’s AI restrictions.
New York Leads Nation in AI Safety
Fifteen states have now adopted chatbot safety laws, with ten passing such measures in 2026 alone. Yet New York’s statute stands out as the most far-reaching. At the federal level, proposals like the GUARD Act and CHATBOT Act remain stalled in Congress, leaving states to address the regulatory void. Data from Common Sense Media indicates that 72% of U.S. teens have used AI companions, with over half engaging with them monthly, emphasizing the law’s broad impact.
Operators now face a 65-day window to assess whether their products qualify for exemptions or require age verification adjustments. The Attorney General’s office must finalize regulations and launch a complaint portal by the same deadline, creating a tight timeline for both industry and regulators. Initially, the law’s effects will be felt by New York residents—both adults and minors—as age verification systems and restricted features take effect on January 1.
