From MediaNama: 2024-05-08 07:58:08

OpenAI is developing new provenance methods to verify content generated by AI. They include tamper-proof watermarking and detection classifiers. Image detection classifier testers found 98% accuracy with DALL·E 3 images. OpenAI has also integrated audio watermarking into Voice Engine. The company has joined C2PA’s steering committee to address content authentication concerns with deepfakes.



Read more at MediaNama: OpenAI is working on new tools to detect AI-generated images