Microsoft has launched an AI Transparency Initiative with more than 40 organisations, including Accenture, Adobe, KPMG, Mastercard and Volvo Cars. The stated goal is a common, open approach to transparency, security and governance across models and providers, addressing the difficulty of reconstructing what happened between an instruction and an agent's outcome.
The initiative separates the problem into seeing, verifying and controlling. Organisations should know which model and runtime were used, what data and tools were accessed, and what actions were taken. They should be able to compare model-reported signals with evidence from independent identity, policy and security systems. They should also retain control of delegated authority and preserve evidence when models or providers change.
Microsoft says it will expand its work in NIST's SI Agent Standards Initiative, collaborate on open standards and bring the resulting evidence into Microsoft AI, Foundry and Copilot. That is a direction of travel, not yet a finished interoperability standard or assurance scheme. The test will be whether competing providers agree on evidence that is sufficiently precise for audits and incident investigations, and whether customers can export and validate it without being locked to one platform's control plane.
