Consumer Reports sent a letter to California Governor Gavin Newsom requesting that he sign the California AI Transparency Act (AB 853). This bill takes on a growing problem for consumers: differentiating between authentic and AI-generated content online.
Last year, California passed SB 942 to ensure provenance information will be embedded into AI-generated content that will allow users to identify its origins. AB 853 complements this effort by:
- Requiring that large online platforms, such as social media sites, mass messaging platforms, and search engines, provide consumers with an easy, conspicuous way to discover if there’s any provenance information available that reliably indicates whether the content was generated with (or substantially altered by) a generative AI system or an authentic content capture device. If that information is available, the large online platform shall make clear the name of the generative AI system, or the name of the device, among other information.
- Prohibiting platforms and websites that make source code or model weights available for download from knowingly making available a GenAI system that doesn’t provide the disclosures required under SB 942. That law requires providers of certain GenAI systems to include latent disclosures in the content their system generates, including the name of the company, the name and version of the GenAI system that created or altered the content, and more.
- At the point of content creation, AB 853 enables provenance markings on authentic, human-generated content by requiring that recording devices sold in California, such as cameras and video cameras, include the option to embed such information.
AI voice and likeness cloning tools have unlocked scammers’ abilities to generate deepfake videos falsely depicting celebrities and political figures endorsing products, suggesting investments, and urging citizens to take action. Recent research suggests that consumers struggle to recognize deepfake videos as false, and also overestimate their own ability to detect deepfakes.
AI-powered celeb-bait has proliferated on social media. An investigation by ProPublica identified videos on Meta seemingly depicting President Trump and former president Biden—each with their distinctive tone and cadence—offering cash handouts if people filled out an online form. 404 Media has reported on the spread of AI clones of Joe Rogan, Taylor Swift, Ice Cube, Andrew Tate, Oprah, and The Rock pushing Medicare and Medicaid-related scams on YouTube. Scammers have used AI deepfake videos of Taylor Swift to hawk Le Creuset dishware. Elon Musk’s likeness has been frequently repurposed by scammers using AI video and voice tools to push fraudulent “investment” schemes. One consumer was reportedly scammed out of $690,000 after seeing a deepfaked Elon Musk endorse an investment opportunity.