The Rise of the "AI Coworker & Assistant" and Why Your Security Isn't Ready

The emergence of Agentic AI, characterized by autonomous and goal-oriented software agents capable of independent decision-making and executing multi-step tasks, is revolutionizing enterprise productivity. However, a 2025 EMA survey conducted on behalf of Ory highlights a significant security gap, as many organizations are rapidly adopting these AI agents without proper security infrastructure in place. While a majority of large and medium-sized companies have deployed AI agents, they often lack documented policies to govern them, creating vulnerabilities due to insufficient identity and access management systems. Human oversight, relied upon by 66% of organizations, is insufficient to manage these agents effectively, as their actions can outpace human monitoring capabilities. EMA suggests treating AI agents as "first-class citizens" by managing their lifecycle with rigorous identity and access controls, enforcing zero trust principles, and ensuring unified visibility for both human and agent activities. Ory addresses these challenges by providing a transparent, modular identity infrastructure solution that supports deployment flexibility and scalability, essential for managing the growing number of AI agents in enterprise environments.