Researcher Details Stealthy Multi-Turn Prompt Exploit Bypassing AI Safety. Well-timed nudges are enough to derail a large language model and use it for nefarious purposes, researchers have found. Dubbed Echo Chamber, the exploit uses a chain of subtle prompts to bypass existing safety guardrails by manipulating the model’s emotional tone and contextual assumptions.
First seen on govinfosecurity.com
Jump to article: www.govinfosecurity.com/llms-tricked-by-echo-chamber-attack-in-jailbreak-tactic-a-28802
