Generative AI refers to AI systems that generate new content, text, images, video, or audio, based on patterns learned from training data. Major generative AI systems include large language models (LLMs), image generation models (diffusion models, GANs), and video generation systems. When generative AI is used without consent to create intimate content depicting real, identifiable individuals, it raises concerns regardless of which specific AI system was used. The commercial generative AI providers (OpenAI, Google, Anthropic, Stability AI, Midjourney) all have safety filters that aim to prevent NCII generation. Perpetrators typically use open-source, locally run models that lack these safety measures.

Key facts about this term

  1. Understand that any AI-generated intimate content raises concerns The implications exist regardless of which AI model generated it.
  2. Use ScanErase's biometric scan to detect AI-generated content Facial geometry matching detects AI-generated content even when it looks different from source photos.
  3. Report NCII-generating AI tools to the FTC AI tools or services specifically marketed for NCII generation may violate FTC unfair practices rules.

Frequently asked questions

Do commercial AI tools like ChatGPT or DALL-E have protections against NCII generation?

Yes. Major commercial AI providers have safety filters that reject requests to generate NCII. Perpetrators typically use open-source, locally run models that have had safety filters removed.

Is using generative AI to create intimate content of a real person a concern even if the content is never shared?

Creation of NCII may violate state laws even without distribution. Distribution raises significant risks under various regulations.

Related glossary terms