I know it’s popular to hate on OpenAI in these forums, but they’re not joking when they say it’s “trivial” to circumvent. They’re also correct that the existence of a watermark doesn’t solve any of the problems you might want it to solve, and it creates some new problems along the way.
Unfortunately, there isn’t going to be a silver bullet that will let you detect automated bullshit and you should stop hoping for one. That ship has sailed.
