USA Law and Practice Contributed by: Lisa Fontenot, Jennifer Broder, Per Chilstrom and Lothar Determann, Baker McKenzie
8.3 Liability Exposures Arising From AI Use AI developers have been subject to class action law - suits and government actions for allegedly: • violating copyrights, privacy laws and computer interference laws for scraping and copying data for development purposes without required permis - sions; • misleading customers and investors about AI capabilities and risks; • discriminating against job applicants with biased systems; and • launching defective products that induced vio - lence, suicides and addiction. Officers and directors are not normally personally lia - ble for violations by their companies, unless they are directly involved in wrongdoing or completely neglect their supervisory duties. 8.4 Key Disclosure Requirements for AI Use Large AI developers must publish transparency reports and disclose critical safety incident reports under California’s Transparency in Frontier Artificial Intelligence Act and other state laws. Companies must also disclose certain details about AI-powered chat - bots to consumers under state laws. Job applicants and employees must be informed about automated decision-making tools under a New York City ordi - nance and other state laws.
• using third parties to assess the potential for cata - strophic risks and the effectiveness of mitigations of catastrophic risks; • revisiting and updating the frontier AI framework; • cybersecurity practices, to secure unreleased model weights from unauthorised modification or transfer by internal or external parties; • identifying and responding to critical safety inci - dents; • instituting internal governance practices to ensure the implementation of these processes; and • assessing and managing catastrophic risk resulting from the internal use of its frontier models, includ - ing risks resulting from a frontier model circum - venting oversight mechanisms. Frontier developers must report critical safety inci - dents to the California Office of Emergency Services and, in case of imminent risk of death or serious physi - cal injury, also within 24 hours to appropriate authori - ties, including law enforcement or public safety agen - cies with jurisdiction, depending on the nature of that incident. Frontier developers must also publish trans - parency reports, disclosing the steps taken to fulfil the requirements of the frontier AI framework, intended uses, restrictions or conditions on uses, results of assessments of catastrophic risks, and other pre - scribed information. Large frontier developers must additionally submit risk assessments to the California Office of Emergency Services every three months. Whistle-blowers are specifically protected by Cal. Labor Code §1107.1, which the California Legislature added via TFAIA. Frontier developers must not pre - vent employees from disclosing violations of TFAIA or certain risks and dangers to federal or state authori - ties, and must notify employees of their whistle-blow - ing rights. Large frontier developers must implement specific whistle-blower processes, and share disclo - sures and responses with officers and directors on a quarterly basis (excluding officers and directors who are specifically the subject of whistle-blower reports).
731 CHAMBERS.COM
Powered by FlippingBook