I would be very, very careful with AI tools meant to detect AI. AI can glitch out in some pretty wild ways, and worse, can be hacked. There's an entire field, adversarial attacks, devoted to using AI to find ways to fool other AI systems. Since the way that generative AI works is that it uses two systems, the disriminator that tries to identify what things are, and the generative system that tries to make things that fool the discriminator, generative AI is already part way there.

Even worse, assuming the claims made in Not with a Bug but with a Sticker are accurate, many of these exploitable glitches are features of most if not all AI systems that are designed to do similar tasks, so once found, they will fool everything that is not specifically trained not to fall for that specific trick. Add to this that I haven't seen any conclusive reason why what YouTube's enegineers called "The Inversion", the moment where AI systems, facing a flood of fake material being designed intentionally to fool them, start to conclude that the human activity is fake, and I've come to think that there's a very real risk that the use of AI tools to detect AI will backfire spectacularly because it will eventually start to determine most human activities are actually AI.