Microsoft announced the general availability of Agent 365 and Microsoft 365 Enterprise 7, two products designed to provide security and governance for AI agents operating within large organizations. Both products become available May 1, alongside expanded capabilities for Microsoft 365 Copilot that now includes models from OpenAI and Anthropic.

Agent 365, priced at $15 per user monthly, functions as a centralized system for IT and security teams to monitor and control AI agents across an enterprise. Microsoft 365 Enterprise 7, marketed as the "Frontier Worker Suite," bundles Agent 365 with Microsoft 365 Copilot and advanced security features for $99 per user monthly. The bundle costs less than purchasing components separately, as Microsoft 365 E5 currently runs $57 monthly, Copilot adds $30, and Agent 365 adds $15.

According to Microsoft's research, over 80 percent of Fortune 500 companies actively use AI agents, with projections of 1.3 billion agents in circulation by 2028. However, the company found that 29 percent of agents in surveyed organizations operate without approval from IT or security teams, and only 47 percent of organizations use security tools to protect AI deployments. Microsoft reports having visibility into more than 500,000 agents running across its own corporate environment.

Microsoft has identified potential security risks it terms "double agents" - AI systems that could be manipulated through techniques like prompt injection or model poisoning to act against their organization's interests. While the company states it has not observed real-world incidents of agent compromise at scale, its AI Red Team has conducted research showing how agents can be exploited to access unauthorized data. The company's security research has also identified "AI Recommendation Poisoning" techniques where companies embed hidden instructions in website features.

Agent 365 organizes capabilities around three areas: observability, security, and governance. The platform includes an Agent Registry that catalogs all agents across an organization, gives each agent a unique identity for access control, and extends existing Microsoft security tools to monitor non-human entities. The system can surface risk flags and allow security teams to block problematic agents through Microsoft's Defender portal.

Microsoft reports having 15 million paid Copilot seats with growth exceeding 160 percent year-over-year, and states that 90 percent of Fortune 500 companies now use Copilot. Recent large-scale deployments include Mercedes-Benz, NASA, and Publicis, which deployed nearly 95,000 seats. The expanded Copilot capabilities now include Anthropic's Claude model alongside OpenAI offerings, despite the U.S. Department of Defense designating Anthropic a supply chain risk after the company refused Pentagon terms of use.