Unsafe Image
Unsafe image research focuses on mitigating the generation and spread of harmful or inappropriate visual content produced by AI models, particularly text-to-image generators. Current research emphasizes developing methods to detect and remove unsafe content, including techniques like prompt purification, feature suppression, and adversarial attack defense, often leveraging large language models and diffusion models. This field is crucial for responsible AI development, aiming to improve the safety and ethical implications of AI-generated imagery across various applications, from social media moderation to preventing the creation and dissemination of harmful propaganda.
23papers
Papers
February 28, 2025
February 25, 2025
February 11, 2025
November 27, 2024
September 22, 2024
March 24, 2024