In the near future, I intend to introduce legislation prohibiting artificial intelligence companies from knowingly generating, facilitating, or distributing child sexual abuse material in Pennsylvania.
The recent federal appellate decision in
United States v. Anderegg exposed a troubling challenge presented by rapidly advancing generative artificial intelligence. In that case, the Seventh Circuit held that the federal government could not punish the mere in-home possession of obscene, wholly AI-generated material when no actual child was depicted. Although the decision is not binding in Pennsylvania, it relied heavily on existing United States Supreme Court precedent.
Importantly, the court’s ruling was narrow. It did not invalidate laws prohibiting the production, distribution or transfer of this material. Whatever constitutional limits may apply to punishing private possession, companies should not be permitted to design, market or operate commercial products that knowingly create and deliver depictions of the sexual abuse of children.
Pennsylvania took an important first step through
Act 125 of 2024, which added artificially generated child sexual abuse material to the Commonwealth’s criminal laws. However, current law provides an exception for an artificial intelligence developer that reports the material after obtaining actual knowledge of its existence. Reporting is essential, but reporting after a system has generated prohibited material is not a substitute for taking reasonable steps to prevent its creation.
My legislation would establish a company-focused framework that would:
- Prohibit AI developers and operators from knowingly generating or distributing material involving an actual child, a sexual deepfake of an identifiable minor or a wholly synthetic depiction that meets the legal definition of obscenity.
- Prohibit companies from intentionally designing, modifying or marketing an AI system primarily for the creation of such material.
- Prohibit the knowing acquisition or use of actual child sexual abuse material to train an artificial intelligence system.
- Require companies that discover prohibited material to promptly disable access, preserve relevant evidence securely and report it through the existing National Center for Missing and Exploited Children process.
- Limit the existing developer exception to companies that maintained reasonable safeguards and responded promptly to inadvertent generation. The exception would not protect intentional or reckless conduct.
- Authorize the Attorney General to seek injunctions and civil penalties against noncompliant companies, while reserving criminal penalties for knowing or intentional violations.
- Preserve appropriate exceptions for law enforcement, judicial proceedings and bona fide child-safety research.
The legislation would not impose strict liability for an isolated attempt to circumvent a company’s safeguards, nor would it require indiscriminate monitoring of private communications. It would hold accountable the companies that knowingly enable this abuse or fail to take reasonable action after discovering that their systems are producing it.
No company should create, nor profit from, technology designed or operated to depict the sexual abuse of children. As artificial intelligence becomes more powerful, the responsibility of the companies developing and deploying it must grow accordingly.
Please join me in cosponsoring this important child-protection legislation.