DALL・E 3 is an artificial intelligence system that takes a text prompt as input and produces a new image as output. DALL・E 3 is based on DALL・E 2 with improved caption fidelity and image quality. This system card shares the work done to prepare for DALL·E 3 deployment. This also includes red teaming external experts, assessing key risks, and mitigating the risks posed by the model and mitigating undesirable behavior.
