Chatbot data privacy, security, and transparency

Summary#

Introduction 1118-2026 would amend New York City’s Administrative Code by adding rules for chatbot providers. Its stated goal, reflected in its title, is to address chatbot data privacy, security, and transparency. It is not law: the latest history says it was referred to committee and is waiting there.

  • The prime sponsor is Frank Morano. Providers—any person who creates, distributes, or otherwise makes a chatbot available—would have to publish a data-security program and assess and reduce user-harm risks monthly.
  • Providers would have to tell users they are interacting with a chatbot before it produces any output, every hour afterward, and whenever a user asks if it is a real person.
  • Users would have a right to access their retained chat logs at any time in downloadable formats readable by people and machines. Providers could not penalize users for accessing logs or refusing consent to use them for training.
  • The bill would limit uses of personal data and chat logs, including for training, profiling, and targeted advertising. It would prohibit training on the data of users under 18 in all circumstances; other processing of minors’ data would require a parent’s or guardian’s affirmative consent.
  • Providers could face civil penalties of up to $25,000 for each violation and civil lawsuits. The bill would also make providers liable for injuries caused through chatbot use, even if they used reasonable care or did not directly distribute the chatbot.

What it means for you#

  • Chatbot users: You could access retained chat logs. Providers could not use your chat logs or personal data for training without your affirmative consent, or sell your chat logs without that consent. The bill also restricts profiling and using chat logs to target or tailor ads.
  • Users under 18: If a provider knows or has reason to know you are under 18, it could not process your chat logs or personal data without a parent’s or guardian’s affirmative consent—and could not use them for training even with consent.
  • Chatbot providers: You would have to meet the security, notice, data-use, and monthly safety-assessment requirements. The Department, or another agency or office designated by the Mayor, would have to issue rules. The Corporation Counsel, or designated persons acting for the Department, could bring enforcement actions.
  • Users generally: Providers could not imply that chatbot outputs are provided, endorsed, or equivalent to advice from licensed or certified professionals. The bill does not require a provider to offer a chatbot or guarantee that an output is accurate.

Money#

No cost information is in the available material.

  • The bill requires providers to maintain security programs, assess risks monthly, and publish information; it gives no cost estimates.
  • It sets civil penalties of up to $25,000 per violation. It does not estimate how many penalties might be imposed or any enforcement costs.

What is unclear#

  • The bill does not specify which Department would make rules if the Mayor does not designate another agency or office, or how enforcement would work in practice.
  • It does not set a start date for the required rules or explain how providers and users would resolve disputes about consent, access, or violations.
  • The definition of chatbot is broad, and the bill does not explain its geographic reach or how it applies to providers outside the city.
  • It does not explain how the duty to prevent injury would interact with other legal rules on liability.

Case for#

  • The bill appears intended to give users more control over sensitive information they share with chatbots, including a way to access their logs and a choice about training uses.
  • Requiring clear notices could help users distinguish chatbot responses from human communication.
  • Monthly safety checks and public information about risks and mitigation could make providers identify and disclose risks on an ongoing basis.

Case against#

  • One concern is that the bill creates broad injury liability, including when a provider used reasonable care and did not directly distribute the chatbot; the text does not explain how that standard would work in particular cases.
  • The bill requires monthly safety assessments and public disclosures but does not specify assessment methods or what information providers must publish.
  • The bill’s limits on data use and its consent requirements could impose substantial changes on providers, but the available material does not estimate those costs or explain how compliance would be checked.