California Enacts New AI Transparency Law Amid Rising Deepfake Concerns
Sacramento, California - October 2023:
In a significant move to combat the challenges posed by rapidly advancing Artificial Intelligence (AI) technology, California Governor has signed Senate Bill (SB) 942 into law. This new legislation is set to offer greater transparency and safeguards against the proliferation of misleading and fake content generated through AI.
AI-Generated Content and Deepfake Concerns
The widespread use of AI, particularly generative AI, has ushered in an era where creating fake audio recordings, videos, and images has become alarmingly easy. These deepfakes are often nearly indistinguishable from genuine content, raising serious concerns regarding authenticity and truth in digital media. Recognising the need for robust measures, the California legislature introduced SB 942, which mandates specific obligations for businesses involved in generative AI.
What SB 942 Entails
SB 942 primarily targets "covered providers", defined as businesses that develop or produce generative AI systems. The stipulation specifies that any provider with over 1,000,000 monthly users or visitors, and is publicly accessible within California, falls under this category.
Key provisions of the bill include:
AI Detection Tools: Covered providers are required to make accessible, at no cost, tools that can detect whether content has been generated or altered by AI. These tools should also provide system provenance data, which details the origin of the data while ensuring no personal information is disclosed.
Content Disclosures: AI-generated content must include clear and conspicuous disclosures indicating it as such. Additionally, latent disclosures – information embedded within the content itself – must be detectable by the AI detection tools to verify the content’s origin and authenticity.
SB 942 in Practice: Case Example from Pikesville
The urgency and potential impact of SB 942 are underscored by past incidents, such as one in Pikesville, Maryland. A high school community experienced significant distress when a recording allegedly featuring racially insensitive and antisemitic remarks by the principal surfaced. It took several months for authorities to confirm the recording was a fabricated deepfake. During this period, the school community faced heightened security concerns, and the principal's professional reputation was jeopardised. Proponents argue that, had SB 942 been in effect, the resulting AI detection tools might have swiftly identified the recording as false, preventing much of the subsequent turmoil.
Enforcement and Penalties
The enforcement of SB 942 is structured to ensure compliance through civil penalties. Violations incur a fine of $5,000 per infraction, enforceable by the Attorney General, city attorneys, or county counsels. This provision emphasises the importance of adherence to the new regulations among AI service providers.
Looking Forward: Implementation by 2026
The requirements set forth by SB 942 will take effect on January 1, 2026, providing a timeline for businesses to develop and implement the necessary AI detection tools. Fortunately, several technologies and standards, like those from the Coalition for Content Provenance and Authenticity (C2PA), are available to assist organisations in embedding and verifying content provenance.
Conclusion
With the enactment of SB 942, California is taking crucial steps to address the challenges posed by AI-generated content. By mandating transparent identification and providing tools for verification, the state aims to mitigate the risks associated with deepfakes and enhance trust in digital media. As the deadline for compliance approaches, businesses are expected to align with these requirements to ensure a more transparent and secure digital landscape.
Source: Noah Wire Services