Just as your LLM once again goes off the rails, Cisco, Nvidia are at the door smiling

Nvidia introduced new NeMo Guardrails Inference Microservices (content safety, topic control, and jailbreak detection) and the Garak tool to evaluate AI vulnerabilities, aiming to keep chatbots and agents aligned and resilient against prompt injection and unsafe outputs. Cisco announced its AI Defense initiative to validate LLM risks, discover shadow AI apps, enforce AI access controls, and unify its own AI agents, with features rolling into Security Cloud and Secure Access. The piece also notes Google’s Titans architecture research and the FTC’s referral of Snap’s MyAI to the DOJ.