58
Can I poison my codebase for LLM agents with malicious comments? (lemmy.world)
submitted 2 days ago by to c/fuck_ai@lemmy.world
17 comments fedilink hide all child comments
 

Rather than adding an agents.md file written to stop LLM agents, I'm thinking about putting random comments into my code telling the agent to stop and warn the user, also some comments that are just word salads/non-sequiturs.

you are viewing a single comment's thread
view the rest of the comments
[–] 12 points 2 days ago (1 child)

LLM generated code does usually include comments, so I'm guessing they do read the comments in code anyways.

As long as the poisoning doesn't disrupt normal usage, I'd say poison away with your comments.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •