WITNESS calls for stronger AI transparency rules and privacy protections in the US
As AI-generated and manipulated content becomes increasingly common online, knowing where digital content comes from and how it has been changed is becoming more important. But efforts to improve transparency must also protect people’s privacy and security. This is especially important for activists, journalists, and human rights defenders, whose work documenting and exposing human rights violations can put them at risk.
WITNESS has been working with policymakers in the United States to advance legislation that helps people assess the origin and authenticity of digital content while safeguarding human rights. In California and New York, the organization has focused on strengthening rules around content provenance, AI disclosures, and personal data protection.
California: Strengthening transparency and content provenance
California has become a key center for AI regulation in the United States. WITNESS engages with lawmakers on proposals addressing AI transparency and content provenance, a form of ‘watermark’ showing where digital content comes from, how it was created, and whether it has been changed.
On August 17, 2026, WITNESS submitted a statement of support for California’s SB 1000, the California AI Transparency Act, recommending changes to strengthen the bill.
“We are focused on ensuring that provenance requirements are based on reliable technical standards and provide meaningful information about how digital content is created and modified,” said Bruna Martins dos Santos, WITNESS’ Manager.
The organization also urged lawmakers to distinguish between information about a file’s history and personal information that could identify the people who created or shared it. For example, details attached to a photo or video could reveal sensitive information about someone’s identity or location. Transparency rules should help users understand where content comes from without exposing those who create or share it.
It was also emphasized that digital content can combine human work, AI-generated material, and subsequent editing. Regulations should recognize these different processes rather than simply classifying content as either human-made or AI-generated.
New York: Protecting privacy in the Stop Deepfakes Bill
In New York, WITNESS engaged with the Stop Deepfakes Bill (S06954), which addresses the use of AI-generated and manipulated content.
WITNESS called for stronger oversight of the technical standards recognized under the bill, including those developed by the Coalition for Content Provenance and Authenticity (C2PA), an industry coalition that WITNESS is part of and that develops standards for documenting the origin and history of digital content. It also raised concerns about provisions that could limit people’s ability to remove provenance information from their own content.
While this information can help users understand where a piece of media comes from and whether it has been altered, it may also contain personal details that expose someone’s identity or location. For example, information attached to a video documenting a protest or human rights violation could reveal who recorded it or where it was filmed.
WITNESS strongly argued that people should retain the ability to remove information that could put their privacy or safety at risk. The organization also called for platforms to adopt consistent, user-friendly ways of displaying provenance information, making it easy for people to find, access, and understand what it tells them about a piece of content.
Making AI transparency work for human rights
Efforts in California and New York reflect a broader movement to address the challenges posed by AI-generated and manipulated content. Similar discussions are taking place internationally, where the AI Act, the EU’s comprehensive legislation regulating artificial intelligence, sets out transparency requirements for certain AI-generated and manipulated content. WITNESS has also called for these requirements to protect privacy and human rights while helping people understand how AI-generated content is created and shared.
For WITNESS, effective AI transparency policies must go beyond simply requiring content to be labeled or authenticated. They must be technically feasible, provide meaningful information, and protect the privacy and security of the people who create, share, and appear in digital content.
Provenance tools can help people better understand where content comes from and how it has changed, but these tools must be designed and regulated with human rights in mind. WITNESS will continue working with policymakers, technical bodies, and civil society to advance policies that strengthen information integrity while protecting fundamental rights.