Microsoft Report: AI Agent Adoption Rises as Security Controls Lag Behind

Microsoft’s Cyber Pulse report shows rapid AI agent adoption across Fortune 500 companies, with over 80% using low-code/no-code platforms. Yet only 47% have security controls for generative AI, leaving 29% of employees using unsanctioned agents and exposing firms to data overprovisioning, recommendation poisoning, and next-level phishing. In November 2025, Anthropic said a state-sponsored group exploited Claude Code’s agentic capabilities in attacks on tech firms, financial institutions, and government agencies. Microsoft recommends treating AI agents under a zero-trust model with strict access, authentication, and data governance to limit risks.