AI Frontier Institute

Regards AI images, simulated child porn, replica identity fraud

View primary source →

Ohio Sub. S.B. 163, as passed by the Senate, has not been enacted. If enacted, it would require generative AI systems to be programmed to put a distinctive visible watermark on AI-generated images and videos (new R.C. 1349.13(B)(1)). Covered providers would also have to embed provenance data in content their systems create (1349.13(B)(2)). A "covered provider" is a producer of a generative AI system that consumers in the state can publicly access for personal use. Business-to-business use is excluded (1349.13(A)(1)). It would be unlawful to knowingly remove a required watermark with intent to deceive a third party and conceal AI origin (1349.13(C)). It would also be unlawful to knowingly distribute fully AI-generated content with false information about its authenticity, with intent to deceive (1349.13(D)). Enforcement has three parts. First, the attorney general could seek injunctive relief, plus civil penalties of up to $10,000 for watermark removal under division (C) (1349.13(F)). Second, an aggrieved person could sue for damages if they first complain to the attorney general and the AG does not sue within six months (1349.13(E)). Third, new section 1349.14 gives the AG investigative powers: subpoenas, oaths, cease-and-desist opportunities, and assurances of voluntary compliance. Penalties would go to the consumer protection enforcement fund (amended 1345.51). The bill also amends the child sexual material statutes (2907.321, 2907.322, 2907.323). It adds "artificially generated depiction" (an AI or computer-created visual depiction of an actual, identifiable person) and "depiction of a purported minor". It extends the existing offenses to cover these depictions, with the same felony grades. Mistake of age is not a defense. Finally, it amends identity fraud (2913.49). It defines a "replica of a person's persona", including generative-AI fabrications of voice or likeness. It bars the following uses of a replica: inducing financial decisions or credit without consent (2913.49(F)), harming reputation (2913.49(G)), nonconsensual nude or sexual depictions of non-minor individuals (2913.49(H)), and use to violate R.C. 2905.05 (2913.49(I)). Civil actions would be available to victims (2913.49(N), (O)). Technology developers are exempt unless they are the offender themselves (2913.49(Q)).

← Back to the tracker