Okta, Inc. has introduced Cross App Access, a new protocol aimed at securing AI agents within enterprises and enhancing visibility and control over app-to-app interactions.

This platform addresses the challenges posed by the Model Context Protocol (MCP), which facilitates connections between AI models and enterprise data while requiring user consent for each integration.

By streamlining the authorization process, Cross App Access is set to improve user experience and mitigate security risks linked to fragmented access controls and token sprawl.

Historically, users were required to manually log in and consent for each app integration, creating security blind spots and inconsistent processes for IT and security teams.

Current integration practices often lack oversight, necessitating manual user consent for each connection and leading to unmanaged security perimeters.

A recent survey indicated that 82% of executives recognize an urgent need for robust AI governance solutions to address risks associated with AI integrations.

Arnab Bose, Okta's Chief Product Officer, pointed out that traditional methods like long-lived tokens and fragmented access controls pose significant risks and hinder AI tool adoption in enterprises.

Bose highlighted that the lack of oversight in app-to-app connections can lead to potential security blind spots and unmanaged perimeters.

The new Cross App Access process eliminates the need for users to manually authorize each integration, thereby enhancing visibility for IT departments by logging all interactions.

This innovative approach aims to reduce security risks, simplify integration processes, and bolster compliance and governance for customers.

Bose emphasized the necessity for improved oversight due to the increasing complexity of AI agents, which introduce unpredictable access patterns and interact with sensitive data.

Current security measures are deemed inadequate for managing the unpredictable behaviors of AI agents, which can compromise enterprise security.