

There was a dark web site kinda like this (just a regular assassination market; not sure if it was real or a honey pot): https://en.m.wikipedia.org/wiki/Assassination_market
It’s also the reason prediction markets have been so heavily regulated (until recently), because they can easily incentivize assassinations while adding plausible deniability.


Some of the “open” models seem to have augmented their training data with OpenAI and Anthropic requests (I. E. they sometimes say they’re ChatGPT or Claude). I guess that may be considered piracy. There are a lot of customer service bots that just hook into OpenAI APIs and don’t have a lot of guardrails, so you can do stuff like ask a car dealership’s customer service to write you Python code. Actual piracy would require someone leaking the model.