SynthID watermark is hard to break, but it doesn't solve AI disinformation

Swap out "AI" -> "Photoshop" and swap "disinformation" -> "hoaxes"

These problems have been around forever. Instead of trying to detect AI, I really think it's more effective to try to use these tools to make a convincing argument why they are false.