They could probably just visit Reddit, there are ample prompts there.
Prompt injection or prompt attacks are well known and likely impossible to guard against. Can you really get a human to be invulnerable to manipulation? Why would we expect the machines to be any better?
Prompt injection or prompt attacks are well known and likely impossible to guard against. Can you really get a human to be invulnerable to manipulation? Why would we expect the machines to be any better?