Synthetic media covers more than deepfake videos. It includes AI-generated voice recordings, text-to-speech systems that mimic someone's voice, entirely AI-generated images, and hybrid content blending real and synthetic elements. In terms of NCII, synthetic media most commonly involves AI-generated intimate images and face-swap videos. The emphasis is on pinpointing any visual representation, whether synthetic or real, that aligns with this definition. Responsibilities for platforms apply to both synthetic and authentic content equally.

Key facts about this term

  1. Synthetic media detection is an evolving technical area Detection companies, academic researchers, and platforms are creating biometric analysis tools, metadata forensics, and model-fingerprinting methods to identify synthetic media.
  2. Future AI technology advancements are part of the conversation Broad language in current laws aims to adapt to future innovations in AI technology without needing new legislation for each emerging modality.
  3. Voice-related synthetic media is a distinct topic Existing laws mainly address visual representations. AI voice clones used for harassment might be dealt with under other legal frameworks (fraud, impersonation, harassment) rather than NCII-specific legislation.

Frequently asked questions

Is a voice clone considered synthetic media?

Yes. AI-generated voice recordings are types of synthetic media. They do not fall under NCII-specific law but could be actionable under impersonation, fraud, or harassment laws.

Which platforms are associated with synthetic intimate media?

All platforms hosting user-generated content and with a substantial number of U.S. users should be vigilant in identifying verified synthetic NCIVD.

Related glossary terms