Teen Pornographers and Trump’s Black Friends: Can Deepfakes Be Controlled?
Technology Policy Brief #109 | By: Mindy Spatt | March 20, 2024
Featured Photo: www.pymnts.com
__________________________________
Regulators and regulators are scrambling to find ways of containing the harms of AI, including the proliferation of deepfakes, which can distort elections and are responsible for a new form of pornography appearing in middle schools around the country. Guidelines passed by the EU, while more than the US has managed, don’t fully address the problem. Some federal legislation has been proposed in the US, but so far hasn’t gone anywhere. A handful of states have taken action.
ANALYSIS
Images of Donald Trump surrounded by adoring black supporters looked to me like obvious fakes. But fake images such as these are not always identifiable and are often being used to influence elections. With concerns over deepfakes growing in the last few years, Google had said it would require political ads using AI to include a prominent disclosure; it is unclear whether the Trump pics were so marked.
According to the Center for Countering Digital Hate, which studied the most popular AI image tools including ChatGPT Plus and Microsoft’s Image Creator, the tools created election disinformation in 41% of cases surveyed, including images that could support completely false claims about candidates or elections.
Voluntary self-regulation is unlikely to solve the problem, but legislators and regulators have been slow to respond to this growing threat. The EU recently approved, with much fanfare, landmark regulations addressing AI, but they don’t ban deepfakes entirely. The rules do require transparency by creators of artificial images in the form of a disclosure of both the origin of the image and the techniques used to create it. Skeptics question whether the EU rules, which haven’t gone into effect yet, will be enforceable and whether they go far enough.
Here in the US Congressman Joe Morelle (D- Rochester, NY) takes aim at a form of AI pornography that often targets young girls with the Preventing Deepfakes of Intimate Images Act. Testifying in support of Morelle’s bill was Francesca Mani, a 15 year old from New Jersey who had fake nude images of her shared online, along with her mother. Her situation is unfortunately not unique. In another high-profile case, five students were expelled from a middle school in Beverly Hills for creating and distributing images of their 8th grade classmates that superimposed actual head shots with AI generated nude bodies.
In introducing the bill, Representative Morelle said “The spread of A.I.-generated and altered images can cause irrevocable emotional, financial, and reputational harm—and unfortunately, women are disproportionately impacted… it’s critical that we take proactive steps to combat the spread of disinformation and protect individuals from compromising situations online.” The Act would make it illegal to share altered or “deepfake” intimate images without consent and makes it clear that consent to create the image does not establish consent for sharing or disclosure. It would also create a right of private action for victims.
An unnamed New Jersey Jane Doe is suing over the same incident Mani testified about, asking a federal district court to stop further distribution of the images and order them destroyed. She is also seeking compensatory damages. This Jane Doe, her lawyers allege, “suffered and will continue to suffer substantial “substantial emotional distress, mental anguish, anxiety, embarrassment, shame, humiliation” and reputational harm caused by the images.
In another case in Florida, two boys, aged 13 and 14, were arrested and face felony charges for using AI to create explicit images of 12 and 13 year old girls. They were charged under a Florida law that makes it a crime to disseminate sexually explicit images without consent. About 10 states have similar laws on the books, but with the criminal charges, possibly the first in the nation, the issue is being taken to a whole new level, and Congress may be spurred to action.
Engagement Resources:
- The New EU AI Act- The Ten Key Things You Need to Know Now, Dentons, December 14, 2023
- Online Safety and Privacy Basics: How to Protect Yourself Against Deepfakes, December 22, 2023
- REPORT: FAKE IMAGE FACTORIES: How AI image generators threaten election integrity and democracy, Center for Digital Privacy March 06, 2024
Check out usrenewnews.org/AI for more news on AI technologies and trends. Stay informed with the latest insights from our dedicated reporters by subscribing to the U.S. Resist Democracy Weekly Newsletter. Your support is crucial in safeguarding fearless, independent journalism. If you appreciate our content, please consider donating today to continue in helping to protect democracy and empower citizenship.