Senate Judiciary Committee hears sponsors on bill to criminalize simulated obscene AI material, require embedded watermarks

The Senate Judiciary Committee held a first hearing on Senate Bill 163, with sponsors Senator Blessing and Senator Johnson testifying that the bill would criminalize simulated obscene material and require AI systems to include a distinctive, embedded watermark to identify AI-generated content.

Senator Blessing, sponsor of SB163, said the bill “seeks to prevent potentially harmful uses of artificial intelligence technology, commonly known as AI.” He told the committee the legislation aims to “protect the safety and privacy of all Ohioans, especially our children, from misuse of this new technology.”

The bill addresses a prosecution gap the sponsors described: current Ohio law criminalizes child sexual abuse material that depicts a real child, but “with AI not being a real photo, this leads to issues of prosecuting someone generating these photos,” Blessing said. He added that SB163 would “give attorneys the ability to prosecute these people.”

Under the bill as presented, AI systems would be required to include a distinctive watermark so that consumers and investigators can distinguish AI-generated products and identify which system produced them. Sponsor testimony described the watermark as potentially embedded in metadata and not necessarily visible to ordinary users: “These are things that can be embedded so that you have no idea that the watermark is actually there,” Blessing said.

SB163 would criminalize making, transmitting, or possessing simulated obscene material that depicts a minor or impaired person in ways that existing Ohio child sexual abuse statutes would prohibit if the person depicted were real. The sponsors also said the bill would criminalize using a replica of a person’s persona — a modified or fabricated voice, photograph, image, likeness, or distinctive appearance — when used to defraud, induce financial decisions or credit, damage reputation, depict the person without consent in sexualized content, or facilitate child enticement or pandering.

Vice Chair Reynolds questioned how broadly a required watermark might apply to benign AI uses such as generative text and asked whether the bill would require a watermark for general AI writing tools. Blessing responded that the bill’s thrust is to target simulated obscene material and related harms, and he said he would examine corner cases raised by the committee.

Ranking Member Hicks Hudson also asked how the watermark would appear; Blessing said it could be embedded in metadata or as a hidden marker not visible to ordinary users and likened it to an invisible but traceable tag useful to investigators.

There was no formal action; committee members recorded the testimony and the hearing was listed as the first hearing on SB163.

The committee transcript records sponsor testimony, follow-up questions from committee members and the sponsors’ explanation of the watermark concept. The sponsors said they want the bill to be workable and indicated willingness to refine definitions and narrow language in response to committee questions.