> At any point in the future someone could sue your project because it turned out the AI had access to code that was copyrighted and you are now on the hook for the damages.

So it is said, but that'd be obvious legal insanity (i.e. hitting accept on a random PR making you legally liable for damages). I'm not a lawyer, but short of a criminal conspiracy to exfiltrate private code under the cover of the LLM, it seems obvious to me that the only person liable in a situation like that is the person responsible for publishing the AI PR. The "agent" isn't a thing, it's just someone's code.