Okta, AWS and CrowdStrike form AI agent security alliance

By Splitfeed EditorialLast updated Sep 26, 2026, 20:18 UTC

On September 22, 2026, Okta announced the Blueprint Alliance together with AWS and CrowdStrike, framing the initiative as an industry effort to build a shared architecture for securing AI agents operating inside enterprise environments. Alongside the alliance announcement, Okta added an AI agent runtime gateway to its own platform, positioning itself as both a convener and a direct infrastructure provider in this space. The headline projection attached to the launch is striking: by 2028, a single Fortune 500 enterprise could be running more than 150,000 AI agents simultaneously.

That number is what gives the alliance its urgency as a topic. If AI agent deployments scale anywhere near that figure inside individual organizations, the identity, access, and threat-detection questions become genuinely difficult. Who or what authorizes an agent to act? How do you audit 150,000 autonomous processes for suspicious behavior? Okta sits in the identity layer, CrowdStrike in threat detection, and AWS in the underlying cloud infrastructure, so the three companies cover complementary parts of what a security model would need to address. What remains open is whether the Blueprint Alliance produces binding technical standards, voluntary guidelines, or something closer to a marketing umbrella for products the companies were already selling.

None of the three camps have published specific reactions to this announcement yet. The Pro-AI camp would typically welcome a major vendor alliance as evidence that the industry is maturing responsibly and that enterprise AI adoption can proceed safely with the right tooling in place. The Anti-AI camp would typically argue that the formation of such an alliance confirms the scale of risk that agentic AI introduces into corporate and critical infrastructure, and would likely question whether voluntary industry coordination is an adequate substitute for regulatory oversight. The Middle Ground camp would generally treat this as a constructive but early step, noting that shared architecture proposals often take years to translate into interoperable, independently verified standards.

The clearest thing to watch is whether the Blueprint Alliance publishes concrete technical specifications or governance commitments, and whether additional major vendors join or notably decline to participate. How the alliance is received by enterprise security buyers and regulators in the months following the September 22 launch will indicate whether this becomes an industry standard or remains one vendor coalition among several.