"AI or Not" is able to detect AI-generated images and audio with 95 percent accuracy, according to developer Optic. Based on a proprietary algorithm, AI or Not can identify images from Midjourney, Stable Diffusion, DALL-E or even classic GANs as AI-generated. Uploaded images are not kept longer than necessary for analysis, according to Optic. However, the team is collecting feedback on the algorithm's performance - presumably to improve it further. In addition to images, the service can also recognize AI-generated music and other audio snippets. AI or Not for images and audio is free for small amounts of images. The company also offers an API that can be used to analyze larger data sets.

Ad
Support our independent, free-access reporting. Any contribution helps and secures our future. Support now:
Bank transfer
Sources
Max is the managing editor of THE DECODER, bringing his background in philosophy to explore questions of consciousness and whether machines truly think or just pretend to.
Join our community
Join the DECODER community on Discord, Reddit or Twitter - we can't wait to meet you.