Open-source Large Language Models can be exploited by Trojan adapters to induce malicious behavior, posing a significant threat to system security and information integrity.
