Hacker News Reader

If you think SynthID-Image can be easily defeated by adding entropy I invite you to give it a try.

I spent half a day messing around with it and I was very impressed by how robust it is. I couldn't get OpenAI to stop detecting their own SynthID without completely trashing the image.

View on HN

Much prior art here, works against both Google and OpenAI's SynthID: https://github.com/0xROOTPLS/DeSynth

Ah you just need to run a local model...

Let me just get my tens of thousands of dollars I have waiting around to get a few sparks.

What about asking LLM to re-draw the image from scratch? Or pass through AI editor?

Anything you generate with ChatGPT has SynthID on it.

The closest thing I found to "defeating" SynthID was to put in a normal photograph and ask ChatGPT to make some utterly trivial edit, and then the output got flagged with SynthID even though it is essentially an unmodified photograph.

There are other LLMs including open-weight ones. They cannot put your ID into the image.

Reply on news.ycombinator.com