This bill creates the “advanced artificial intelligence licensing act,” a comprehensive regulatory framework for advanced AI systems in New York. Part A adds a new article to the state technology law establishing definitions, a Department of State licensing and enforcement structure, an advisory council on artificial intelligence, rulemaking authority, hearing procedures, penalties, and investigative powers. It also amends the criminal procedure law to authorize investigators appointed by the Department of State to have peace officer powers for enforcing the article.
Part B requires registration and licensing of high-risk advanced artificial intelligence systems and imposes ongoing compliance obligations on operators. Covered systems include AI used in healthcare, transportation, finance, government decision-making, surveillance, critical infrastructure, autonomous weapons, and neural or cognitive decoding. Operators must disclose systems, obtain licenses, maintain ethics and risk management boards, submit annual risk reports, preserve logs and backups, seek approval for certain source code modifications or rewrites, and notify the state of malfunctions or incidents. The bill also authorizes the secretary to inspect source code, logs, books, and records, and to suspend or revoke licenses for violations.
Part C establishes a binding ethical code of conduct for licensees and non-licensees developing or operating high-risk AI, emphasizing respect for autonomy, equity, accountability, care, privacy, inclusivity, human oversight, notice, and safety. Part D goes further by prohibiting certain AI systems outright, including systems using subliminal manipulation, causing unjustified harm, making unlawful predictive inferences that infringe liberty or financial interests, unlawfully accessing sensitive data, or operating autonomous weapons without meaningful human control. The bill allows a narrow exception for otherwise prohibited systems only if authorized by the secretary and used by the state or under substantial state oversight after public hearing and comment.
The bill would significantly expand state oversight of AI by creating a licensing regime, mandatory reporting and logging requirements, and enforcement tools that include civil penalties, criminal penalties, injunctions, and administrative hearings. It would affect AI developers, operators, platform hosts, and some third-party integrators, while also giving the Department of State broad authority to regulate source code changes, compliance certifications, and public disclosure of licensees. In practical terms, it would impose substantial compliance obligations on companies deploying AI in sensitive sectors and could restrict or prohibit certain model development and deployment in New York.
The overall sentiment reflected in the bill text is cautious and regulatory, but not uniformly anti-innovation. The findings emphasize both the benefits of AI and the need for “balanced” and “non-intrusive safeguards,” suggesting an intent to curb harmful uses without blocking beneficial innovation. Because there are no committee transcripts or votes provided, there is no recorded public debate or voting history here to indicate broader legislative support or opposition.
The main points of contention likely center on the breadth of the definitions, the scope of state authority, and the treatment of high-risk and prohibited systems. Potential concerns include whether the licensing and inspection regime is too burdensome, whether the secretary’s discretion is too broad, how the bill would affect open-source or publicly accessible code, and whether the criminal penalties and restrictions on source code changes could chill innovation. Supporters would likely emphasize consumer protection, privacy, safety, and accountability, while critics may focus on compliance costs, enforcement reach, and possible conflicts with interstate commerce or federal AI policy.
The bill would amend the state technology law to create a new regulatory article governing advanced artificial intelligence systems, and it would also amend the criminal procedure law to grant Department of State investigators peace officer authority for enforcement purposes. It would establish licensing, registration, reporting, inspection, and penalty provisions for high-risk AI systems, require ethics and risk management boards, mandate logging and incident reporting, and authorize the state to review source code and system outcomes. It would also create a separate prohibition on certain AI systems and allow the state to order cessation of development or operation, suspend licenses, and impose civil and criminal penalties on violators.
The bill’s tone is strongly precautionary and regulatory, reflecting concern about the risks of advanced AI while still acknowledging its potential benefits. The text frames the proposal as a balanced framework intended to prevent harm, preserve public safety, and support responsible innovation. No committee transcript or vote data were provided, so there is no direct evidence of legislative debate, amendments, or recorded support/opposition beyond the bill’s own findings and structure.
The likely points of contention are the bill’s broad definitions of “high-risk” and “prohibited” AI, the Department of State’s wide discretion to require licensing, halt development, inspect source code, and impose penalties, and the compliance burden on developers and operators. Open-source and publicly accessible code may be especially sensitive because the bill treats platform hosts as operators in some cases and allows state intervention even before deployment. Another likely dispute is whether the criminal penalties, mandatory ethics boards, and restrictions on source code modifications are necessary safeguards or overly intrusive measures that could discourage innovation and create legal uncertainty.