Meta Unveils LlamaFirewall: The Open-Source Shield Against AI Vulnerabilities

Meta introduces LlamaFirewall, an open-source framework to protect AI systems against cyber threats with three key guardrails: PromptGuard 2, Agent Alignment Checks, and CodeShield. The modular architecture enables layered defenses for both simple chat models and complex autonomous agents....