OpenAI Adopts Google's SynthID Watermark for AI Images with Verification Tool

OpenAI just made a very public trust move: it says AI-made images from its tools will now carry stronger origin signals, including Google’s SynthID watermark and a preview of a public checker tool so people can test whether an image came from OpenAI. In plain English: the company wants it to be easier to tell when a picture was created by AI, at a time when fake visuals are getting harder to spot and easier to spread.

But the real fireworks were in the comments, where the crowd instantly split into Team Finally, Some Accountability and Team Cute Idea, Won’t Last Five Minutes. The biggest skeptical take came from users who basically said, “Be serious — anyone making propaganda will just remove the label.” That mood popped up again and again, with people wondering whether these marks are just a speed bump for bad actors rather than a real lock on the door. One commenter even imagined a future where platforms punish marked images so aggressively that the watermark gets stripped out “overnight,” which is… not exactly a glowing vote of confidence.

Still, defenders pushed back. One user bluntly noted that lots of people claim these watermarks are easy to defeat, but where’s the proof? And then came the classic internet side-quest: complaints that Google’s system is too closed off, with one commenter practically waving an open-source repo in the air like, “Fine, I’ll build my own.” Add in worries about whether verification requires uploading an image somewhere, and the vibe is clear: people want safer AI, but they also want transparency, privacy, and a system that doesn’t feel like security theater.