New York Attorney General (AG) Letitia James has created a new AI enforcement unit, the AI Safety, Standards, Enforcement, and Transparency (ASSET) Unit, to police how companies develop and use AI. The unit will put data privacy and cybersecurity at the center of its work, and it will lead enforcement of the Responsible AI Safety and Education (RAISE) Act, which takes effect January 1, 2027. Any company that uses AI with New York customers or their data should start preparing now.
Does the ASSET Unit Create New Legal Requirements?
No. The unit does not add legal duties on its own. It will enforce duties already in place and those arriving in 2027. Large AI developers will have new duties on January 1: they must publish their safety measures and report incidents to regulators. Other companies face no new requirement because of the unit itself, but the unit adds technical expertise to enforcement of laws already on the books, including those on chatbot disclosure, data security, and AI in advertising.
What Will the ASSET Unit Do?
The unit will lead the AG’s enforcement of the RAISE Act and other AI safety laws, develop policy proposals, and give technical support to investigations across the office. Its stated focus includes data privacy, digital scams, cybersecurity, and online safety.
Who Will the ASSET Unit Be Watching?
The RAISE Act, which the unit will enforce, requires developers of certain advanced AI models to publicly disclose their safety practices and promptly report serious safety and security concerns. Bloomberg Law reports that developers such as OpenAI, Anthropic, and Meta are expected to register with New York regulators before the January deadline, and that the law applies to any company with $500 million or more in annual revenue. But the unit’s mandate is broader than developers, as the laws below show.
Could the Unit’s Job Grow?
Possibly. State Sen. Andrew Gounardes has introduced a follow-on bill, called “RAISE 2.0,” that would require quarterly reports from large developers, require them to report significant gains in a model’s capability or computing power within 72 hours, and give third-party auditors “employee-level access.”
It would also let a new state office order a developer to slow its computing power. It is a proposal, not law. Under RAISE 2.0, that power would sit with a separate office, the Office of Digital Innovation, Governance, Integrity, and Trust, not the ASSET Unit. How the two offices would divide enforcement is an open question. The timing matters: the legislative session begins in early January, so the ASSET Unit will start work while lawmakers consider more AI bills.
What Laws Can the ASSET Unit Use?
The AG’s office points to these existing and upcoming laws:
- New York’s Stop Hacks and Improve Electronic Data Security (SHIELD) Act: requires reasonable data security to prevent breaches.
- RAISE Act: requires large AI developers to publish safety measures and report incidents. Effective January 1, 2027.
- S.9051B/A.10379C: bans unsafe chatbot features for children. Effective June 2027.
- General Business Law, Article 47: requires AI chatbots to try to detect and address suicidal ideation and to tell users they are not talking to a human.
- Penal Law § 245.15: criminalizes sharing intimate images with intent to harm, including AI-generated or AI-altered images.
- Synthetic Performer Disclosure Act: requires disclosure when ads use AI in place of humans.
The office also has broad authority to act on privacy violations, unfair or deceptive business practices, and hacking, including under the federal Computer Fraud and Abuse Act. A company does not need to be an AI developer to fall within it.
What Should Companies Do Now?
Companies that build, deploy, or advertise with AI in New York should take these steps:
- Know which of the unit’s laws could reach you. The RAISE Act covers large AI developers. Chatbot, AI-in-advertising, and data security laws may reach companies that deploy AI or hold New Yorkers’ personal data.
- Be ready for January 1. When the RAISE Act takes effect, large developers must publish their safety measures and report incidents. Decide what counts as a reportable incident, who decides, and how fast you can report. Make sure public statements match internal practice.
- Tighten data security and AI controls. Review safeguards around training data, customer data, and systems your AI tools can reach. Set clear limits and logging for AI agents. For chatbots, tell users they are not talking to a human, build in self-harm safeguards, and prepare for the June 2027 rules on features for children.
- Make it easy to raise concerns internally. The AG has invited AI industry employees to file whistleblower complaints. Give employees an internal path, and do not discourage them from speaking up.
- Watch the unit’s first moves. Look for new hires, guidance before January 1, and whether early inquiries focus on disclosures, incident reporting, or data security.
