OpenAI launches invisible watermarks to spot AI generated images
Technology
OpenAI just launched a smarter way to spot AI-made images by adding invisible watermarks.
This upgrade comes after earlier attempts, like metadata tags in 2024, didn't really stick. People found ways to get around them.
With these new signals, it should be much easier to tell if an image was created by AI.
OpenAI uses SynthID and public verifier
OpenAI is using SynthID (from Google DeepMind) to embed digital watermarks you can't see but that survive edits like cropping or resizing. The tech meets C2PA standards for authenticity.
Plus, there's now a public tool on OpenAI's website where anyone can check if an image was generated by tools like ChatGPT or Codex, making it simpler for everyone to verify what's real and what's AI-made.