Most States Have Enacted Sexual Deepfake Laws
Key highlights this week:
We’re currently tracking 680 bills in 45 states (plus another 132 congressional bills) related to AI this year, 67 of which have been enacted into law.
Arkansas is the latest state to dedicate a working group to studying AI policy.
The governor of Massachusetts has signed a sexual deepfake bill into law, joining 26 other states, which is the topic of this week’s deep dive.
The AI-related issue that states have acted to address the most quickly and extensively is sexual deepfakes. Currently, 27 states have enacted new laws to address the proliferation of sexual deepfakes generated by AI. Disturbing stories from local high schools and celebrities have spurred lawmakers to action, but these bills are also relatively easy to add to existing revenge porn and child sexual abuse laws.
These images often contain the face of an actual person on a naked or partially clothed body that is not their own with users disproportionately targeting women. Along with deepfakes aimed at electoral candidates, states are moving quickly to combat the alarming proliferation of sexual deepfakes.
Deepfakes were the subject of about half of the over 600 AI-related bills that state lawmakers introduced this session. They also make up over half of the AI-related bills enacted this year. Deepfake images, videos, or audio are an easy, if startling, example of AI-generated content that lawmakers can see and hear with their own eyes and ears. When combined with national headlines and local anecdotes, state policymakers have been pressed to address deepfakes quickly.
Since we last addressed this topic in December, nearly 20 additional states have enacted new laws addressing sexual deepfakes. Another trend that has solidified this year is that these laws are splitting between sexual deepfakes targetting adults and deepfakes dipicting children. The first set of laws is the use of nonconsensual AI-generated images and videos showing real individuals depicted in a sexually explicit manner. These laws date back to 2019, when Virginia became the first state to address sexual deepfakes (VA HB 2678) by adding nonconsensual sexual deepfakes to an existing “revenge porn” law along with California (CA AB 602) later that year. Two years later, Hawaii (HI SB 309) and Georgia (GA SB 78) enacted their own sexual deepfake laws in 2021.
Last year, the trend shifted when Texas (TX SB 1361), New York (NY SB 1042A), and Minnesota (MN HF 1370) added criminal penalties for violation of the nonconsensual deepfake laws (the previous laws were limited to civil lawsuits). And this year, we saw 12 additional states enact new laws to limit nonconsensual, sexual deepfakes.
In 2023, we also saw the first instance of states adding AI-generated content to their laws criminalizing sexually explicit images depicting minors in Louisiana (LA SB 175) and Texas (TX HB 2700). Similar to the nonconsensual sexual deepfake laws, which can be amended to current revenge porn statutes, laws criminilizing deepfake images of minors can be added to current child sexual abuse material (CSAM) statutes.
This year, 12 states enacted laws directly adding AI-generated media to current CSAM statutes. Currently, of the 27 states that have enacted sexual deepfake laws, 13 focus on restricting the distribution of nonconsensual sexual deepfakes, 8 criminalize AI-generated CSAM, and 6 states have addressed both issues.
Most of these laws explicitly add AI-generated media to actions that were already illegal. This is a classic case of conduct-level regulation of AI, a “broadly technology-neutral approach” that utilizes or expands existing laws to enforce against undesirable conduct from AI. And while much of the focus has been on sexual and political deepfakes, we’re seeing bills introduced to combat deepfakes within current fraud statutes as well.
Recent Developments
In the News
Claude Sonnet 3.5: Last week, Antrhopic released its newest AI model, Sonnet 3.5. While Sonnet is technically Anthropic’s middle-tier model, they claim that Sonnet 3.5 performs better than its flagship larger model, Opus 3.0, while maintaining the speed and cost of a mid-tier model. The new model is available for free and, so far, users are impressed.
Major Policy Action
Arkansas: On Wednesday, Gov. Sanders (R) launched a working group to study and offer recommendations for the safe use of AI within state government. Robert McGough, the state’s chief data officer, will lead the working group. The AI & Analytics Center of Excellence will exist for one year, will meet monthly, and will provide an initial report to the governor by Dec. 15, 2024.
Massachusetts: Last Thursday, Gov. Healey (D) signed a sexual deepfake bill (MA H 4744) into law. The new law makes it a crime to distribute visual material, including material produced by digitization, depicting another person nude or in sexual conduct that causes distress with the intent or with reckless disregard of harm and without consent.