Insygna joins Open Secure AI Alliance to push AI agent accountability

Insygna has joined the Open Secure AI Alliance as a general member and The Linux Foundation as a silver member, putting its agent-management tools alongside work from NVIDIA, Microsoft, Okta and others. The move highlights growing pressure to give companies verifiable identity, access control and audit trails for AI agents before those agents spread across the enterprise.

Why it matters: - AI agents are moving from demos into day-to-day work, which creates new questions about access, cost, and accountability. - Insygna's pitch is that companies need the same controls for AI agents that they use for employees and contractors. - The alliance is trying to build those controls as open infrastructure rather than locking them inside one vendor's platform.

What happened: - Insygna said Sept. 28, 2026, that it was accepted as a General Member of the Open Secure AI Alliance. - Insygna also became a Silver Member of The Linux Foundation. - The company said it will join alliance work focused on AI agent identity and control. - Insygna will appear at Staffing Industry Analysts CollaborationX in Dallas on Sept. 29-30, 2026, as a Shark Tank finalist. - Insygna will also exhibit at HR Tech in Las Vegas at Booth SU12 and at TechCrunch Disrupt in San Francisco in October 2026.

The details: - Insygna manages AI agents after deployment, when the operational cost and risk show up. - The platform assigns each agent a verifiable identity. - The platform credentials each agent against a defined scope of authorization. - The platform manages agents from deployment through retirement. - The platform keeps a retained log of agent activity in language a procurement lead or auditor can read. - Insygna also runs the Insygna Agent Report Card, a free public service that scores an AI agent's risk before deployment. - The report card is free to anyone and will remain free. - Michail Rybakov, Insygna cofounder and CTO, said the company scans an agent's codebase to assign a trust score and then tracks its activity continuously. - Rybakov said managers do not need to be security experts to understand what an agent can do or whether it is behaving properly. - Noah Beygelman, AI Security and Compliance Manager at Insygna, said employees are already spinning up agents without real cost controls or company-wide risk management.

Between the lines: - The alliance now includes more than 120 organizations, signaling broad industry interest in shared AI security standards. - Publicly named participants include Microsoft, IBM, Red Hat, Cisco, Cloudflare, CrowdStrike, Okta, Palo Alto Networks, Hugging Face, Databricks, Palantir, Salesforce, SAP, ServiceNow, Siemens, Dell and HPE. - The alliance operates as a directed fund project of The Linux Foundation and was convened by NVIDIA on July 27, 2026, under a charter effective Sept. 1, 2026. - Membership requires an application, review, and an agreement countersigned by The Linux Foundation. - Agent identity is one of the alliance's active workstreams, including contributions from Okta, Palo Alto Networks and HPE's SPIFFE/SPIRE project. - Insygna's angle is different from the infrastructure providers: it focuses on the people who hire, credential, and pay for the agents. - On Aug. 4, 2026, NVIDIA and The Linux Foundation proposed SAFE, a framework for confidential AI security reporting and shared mitigations, and the proposal remains open for public comment.

What's next: - The alliance is expected to keep expanding open work on identity, isolation, model formats, multi-model scanning, and secure coding. - Insygna said it wants more companies to adopt controls that show what AI agents cost, what they can access, and who is accountable for their actions. - Insygna's public report card and enterprise workflow tools position the company to push the same accountability model into HR, Finance, and Procurement teams.

The bottom line: - AI agent security is shifting from a technical problem to an enterprise governance problem. - Insygna is betting that the winning controls will be the ones non-technical managers can actually use.