The evolving landscape of AI agent security

AI agents are increasingly integrated into enterprise operations, performing tasks such as web browsing, coding, and API interactions. This integration brings significant productivity gains but also introduces new security challenges. Companies must now secure not only users and devices but also software actors that act on their behalf. These agents require permissions, monitoring, and governance to ensure authorized actions. As enterprises scale from experimenting with a few agents to deploying hundreds, agent identity becomes a critical cybersecurity layer. Unlike traditional users, agents can move between systems and make decisions, complicating control. The market is unlikely to consolidate into a single category called 'AI security.' Instead, opportunities lie in specific control points. Companies are addressing different aspects, such as agent identity, data access, and traffic security. Kiteworks acquired Bonfy.AI, focusing on real-time data classification, while Huskeys, a cybersecurity startup, raised $27 million to secure complex internet traffic. These developments highlight the emergence of distinct security layers. Entrepreneurs are advised to focus on specific capabilities rather than broad 'AI security' positioning.