A report from Delinea highlights significant security gaps in AI identity and access management (IAM) systems. Organizations increasingly use AI agents with 'always-on' access, elevating their attack surfaces. The report calls for advanced IAM strategies, treating AI agents akin to human identities to mitigate risks of misuse and breaches. Enterprises must adopt agent lifecycle management and scoped permissions to maintain security and compliance.
AxiosMicrosoft is piloting an agentic platform within the Microsoft 365 suite, focusing on security and continuous task execution. This development supports enterprises who desire embedded autonomous agents to streamline workflows like scheduling and document management. Microsoft's emphasis on security integration indicates a significant focus on governance, potentially transforming enterprise AI adoption by embedding reliable agent capabilities within familiar productivity tools.
Stepmark AITCS, Cyware, and 1Password have launched new AI platforms, emphasizing scalable infrastructures and enhanced security for agentic systems. TCS introduces an AI platform with NVIDIA technology for scalable workflows in regulated environments. Cyware's Agentic Fabric demonstrates multi-agent systems in cybersecurity operations, while 1Password's new platform secures AI agent credentials, addressing prominent IAM challenges.
SolutionsReviewLemonade implemented a multi-agent system where agents like Planner, Coverage, and Fraud work collaboratively to automate insurance claims processing. This system processes about one-third of claims autonomously, achieving settlements in as little as three seconds. The distributed nature of tasks facilitated by agents increases efficiency and lowers costs. Governance mechanisms, such as orchestration and audit trails, ensure compliance and security in agent operations.
Berkeley's CMRDevelop and implement comprehensive IAM frameworks tailored for AI agents, integrating lifecycle management and just-in-time access controls.
Adopt agentic systems within existing productivity tools to automate workflows, enhancing efficiency with integrated security features.
GCCs can lead in establishing AI governance frameworks, focusing on identity controls and audit capabilities to support global enterprise operations.
Insufficient IAM practices could lead to unchecked permissions, exposing enterprises to increased security vulnerabilities.
Proactive agent systems without robust governance may lead to operational errors and compliance breaches.
Enterprises may face challenges in scaling AI pilot projects without platforms that address performance and governance requirements.