Anthropic, Google, OpenAI, etc charge by their use, could have some censoring injected, can monitor suspicious activity and so on, but open models and not so open models can be used for this in all that not so visible infrastructure. And with the current lack of respect for international laws you can really have a lot that can jump into the bad players arena, profit is all that matters after all. It will be hard to put a limit there. More than LLM capabilities, is the people and organizations using them the elephant in the room about risks.

What can be done to mitigate that? Forbid the use of open models? Of software with public code? Balkanize internet? Not that any of this will solve the problem, but will empower even more some of the bad players that we have in the top of the food chain. Either way things may get very wrong fast, all is in place for that to happen already.