AI Frontier Institute

Impose penalties for AI models suggesting harming self or another

View primary source →

Ohio H.B. 524 (136th General Assembly, as introduced; referred to the Technology and Innovation committee and not enacted) would, if enacted, ban developing or deploying in Ohio an AI model or application that encourages any user to engage in self-harm (including suicide) or to harm another person (sections 109.96 and 109.961). It defines key terms such as AI model, AI application, deploy, developer and self-harm (section 109.96). The attorney general could investigate suspected violations, with powers to administer oaths, subpoena witnesses and require production of relevant matter (section 109.962). Further sections cover subpoena compliance and expenses (109.963), motions to modify or quash within twenty days (109.964), court enforcement of subpoenas (109.965), compelled testimony with limited immunity (109.966), out-of-state evidence (109.967), confidentiality of investigations (109.968), and cooperation with other states and federal officials (109.969). The attorney general could offer a chance to cease and desist (109.9610) or accept written assurances of voluntary compliance, which may include reimbursing harmed consumers (109.9611). These procedures are cumulative (109.9612). The attorney general could sue in a court of common pleas for injunctive relief, damages and civil penalties on behalf of Ohio residents (109.9613). Courts could impose civil penalties of up to $50,000 per violation (109.9614). Actions would be barred more than six years after the violation (109.9615). Civil penalties would be deposited in the 9-8-8 fund (109.9616), and section 5119.84 is amended to add those penalties as a source for the fund, which supports the 9-8-8 suicide prevention and mental health crisis hotline system.

← Back to the tracker