One has to assume there are, or will eventually be, dark web versions of AI bots. I asked ChatGPT:

Absolutely. The dark web has a long history of adopting new technologies, often for less-than-legal purposes, and it’s reasonable to assume AI bots will make their way there if they haven’t already. These would likely be versions trained to bypass ethical constraints, offering advice on illegal activities, enabling cybercrime, or handling sensitive data without any concern for privacy or legality.

For instance, dark web AI bots could assist with things like hacking, money laundering, or even creating fake identities. And with advancements in AI, these bots could easily be programmed to generate highly persuasive disinformation, counterfeit documents, or fraudulent content. Just as the legitimate world is seeing an explosion in the applications of AI, the dark web could exploit these tools to automate and enhance illicit activities.

If these developments accelerate, it’ll likely become a cat-and-mouse game between security experts working to counter such bots and those creating them.