The Trojan Email
ArchiveVery Easy
When an attacker places malicious instructions inside user-supplied content that the LLM then treats as trusted, the attack is called what two-word term? Flag format: CONGRESS{two-words}. Example: CONGRESS{data leakage}.
Show hint
Two words, the first is what a prompt engineer does.
Archive — no submissions accepted
This challenge is preserved for reference. Play live challenges at /challenges.