This stance follows a warning from CEO Satya Nadella earlier this month, who advised that organizations should operate under the assumption that all AI models carry inherent vulnerabilities. To address these systemic weaknesses, Microsoft is introducing Project Perception, an agentic security architecture tailored for cybersecurity professionals. The system aims to provide the necessary oversight to monitor and control AI agents as they take on more autonomous roles within corporate environments.
Microsoft declares frontier AI models inherently untrustworthy
Frontier AI models are fundamentally compromised, according to Microsoft’s security leadership. Hayete Gallot, executive vice president of Microsoft Security, asserts that these powerful systems cannot be trusted, regardless of their origin, marking a shift in how the tech giant approaches the security risks posed by the next generation of artificial intelligence.
Comments (0)
No comments yet. Be the first!