Microsoft, Nvidia Move Enterprise AI Toward Active Cyber DefenseKey Takeaways Microsoft Project Perception uses autonomous AI agents for machine-speed enterprise cyber defense. Nvidia's Open Secure AI Alliance promotes transparent, auditable, open source AI security tools. Enterprise AI security is shifting from model protection toward active, human-governed defense. AI security is entering a new stage as organizations look beyond protecting models from prompt injection, unauthorized access, and data exposure. "The physics of cybersecurity are changing," Hayete Gallot, executive vice president of Microsoft Security, wrote in the announcement.