Generative AI
AI techniques (e.g., GANs, transformers) that create new content—text, images, or other media—often raising novel governance and IP concerns.
Definition
A family of models that learn data distributions and sample new instances. Generative AI powers text-to-image, deepfake, and code-generation tools. Governance must address data provenance (licensing of training data), content moderation (preventing harmful generations), watermarking (to identify machine-generated media), and intellectual-property safeguards to respect third-party rights and avoid misuse.
Real-World Example
A marketing agency uses a text-to-image generative model to create product mockups. Governance policies require that training images come from licensed stock photos, generated content is watermarked, and a human review board vets all outputs for brand compliance and potential copyright issues.
.svg)
