AI agents do not stay in one place. They call APIs, invoke tools, cross cloud boundaries, and interact with data sitting at the edge of enterprise networks. That distributed behavior is precisely what makes them useful, and precisely what makes governing them genuinely difficult. Ping Identity is addressing that problem directly with new integrations across AWS, Google Cloud, and Cloudflare.
The integrations extend Ping’s Runtime Identity approach into the environments where AI agents actually operate rather than just where they originate. The practical difference matters. Traditional identity and access management focuses on authenticating who is logging in. Runtime Identity extends that logic into continuous authorization, asking not just who an agent is but what it is doing, what it can access, and whether each action aligns with current policy at the moment it happens.
On AWS, the integration covers AI agents built on services including Amazon Bedrock AgentCore, applying least-privilege access controls as agents interact across APIs, tools, and multi-account environments. Hart Rossman, AWS’s VP of Security and Infrastructure, noted that production AI workloads need authorization controls that adapt dynamically as agents move across cloud environments, which is a more honest description of the problem than most vendor announcements tend to offer.
The Google Cloud integration works through Cloud Agent Gateway, where Ping’s PingOne Authorize product enforces fine-grained policies across agent and tool interactions including MCP tools and downstream APIs. Centralizing that authorization logic gives security teams a single point of visibility into which agents are doing what, rather than trying to reconstruct that picture from logs scattered across individual services.
Cloudflare extends the coverage further out to the edge, where agents interact with public and private data across distributed infrastructure. Cloudflare currently runs GPU inference capacity across 220 cities globally, which gives some sense of how geographically dispersed AI agent activity has already become. The joint work with Ping applies Zero Trust policies to agent traffic at that layer, monitoring and auditing activity where it reaches the furthest from the enterprise perimeter.
Andre Durand, Ping’s CEO, described the core tension plainly. Organizations want to move faster with AI, but autonomous agents operating across cloud and edge environments without real-time visibility create governance gaps that security teams cannot afford to ignore.
