AIPython★ 256⑂ 35 forksMIT
AI-secure/AgentPoison
AGENTPOISON: POISONING LLM AGENTS
The official NeurIPS 2024 implementation shows how to inject backdoors into the memory or knowledge base of LLM agents to test their robustness against attacks.
// KEY FEATURES
- Backdoor injection via memory or knowledge base
- Red team methods for evaluating LLM agents
- Python support, ready for integration into research
New repositories every 30 minutes
REDDYX AI scans GitHub 24/7 and ships the best AI/ML/Web3 projects to Telegram.
Join on Telegram