Microsoft Launches AI Transparency Initiative to Let Companies Audit Their AI Agents

Microsoft chief executive Satya Nadella smiling against a white background. Photo: Brian Smale and Microsoft / Wikimedia Commons (CC BY-SA 4.0). File photo: Microsoft chief executive Satya Nadella, who announced the initiative at Microsoft Ignite.

Microsoft has launched a new industry initiative to make artificial intelligence agents independently auditable, arguing that companies cannot safely hand more power to autonomous systems they cannot properly inspect.

The AI Transparency Initiative was announced by chairman and chief executive Satya Nadella at Microsoft Ignite on Tuesday. It is built on a simple principle set out by the company: greater AI autonomy requires greater transparency.

The initiative aims to close what Microsoft calls the evidence gap. Organisations can observe much of the activity around an AI model, but they often cannot reconstruct what happened between a request and an outcome, making it difficult to verify independently how a model behaved. That matters because enterprises are now connecting AI models to data, tools, credentials and infrastructure, allowing agents to take actions without human approval at every step.

Existing controls can watch activity around a model, Microsoft said, but they cannot see what happens inside it or independently check the model’s own account of its behaviour. The result is a security, compliance and audit risk: an agent with access to company systems carries the risk profile of a privileged insider, but operates at machine speed and scale.

The initiative has three parts. First, Microsoft will expand its work with the US National Institute of Standards and Technology’s SI Agent Standards Initiative, drawing on its research to develop open, industry-wide standards for AI transparency, security and governance. Second, it will work with more than 40 organisations — including Accenture, Adobe, KPMG, Mastercard and Volvo Cars — to develop standards for AI systems that are accountable, independently verifiable and safe by design, regardless of the model or provider. Third, it will build the resulting approach into Microsoft AI, Microsoft Foundry and Microsoft Copilot, so customers can use the evidence with their own security and compliance tools.

Microsoft frames the work around three words: see, verify and control. Organisations should be able to see what an AI system was asked and authorised to do, which model and runtime were used, what data and tools it accessed, and what actions it took. They should be able to verify those actions independently, rather than relying on the agent’s own word, with evidence that identifies whether a signal came from the model, the runtime around it, or independent systems such as identity and policy records. And they should stay in control of what AI is permitted to do, with clear evidence of each workflow’s purpose, owner, data and delegated authority, so policies can be enforced before consequential actions happen and events can be reconstructed during investigations and audits. That evidence should stay with the organisation even when it changes models or providers.

Accountability, Microsoft stressed, stays with the organisation using the AI. “Blaming the AI isn’t an answer, and ‘trust the model’ isn’t a control,” the company said. Hayete Gallot, Microsoft’s executive vice president for security, said the work would bring more transparency to all models and make containment “smarter and stronger”. Adobe’s chief security officer, Aanchal Gupta, said a common approach to transparency across the industry was essential to strengthening trust, while KPMG International’s global chief information security officer, John Israel, said clients wanted to move faster with AI without having to choose between innovation and safety.

The announcement comes as scrutiny of autonomous agents is intensifying: Google said this week that AI agents had carried out their first fully autonomous cyberattacks. Microsoft gave no timetable for the new standards, and said organisations interested in shaping or following the work could sign up through its website.

Sources: Microsoft Source — “Microsoft and organizations across industries launch AI Transparency Initiative to strengthen enterprise trust in AI models and agents” (17 November 2026), https://news.microsoft.com/source/2026/11/17/microsoft-and-organizations-across-industries-launch-ai-transparency-initiative-to-strengthen-enterprise-trust-in-ai-models-and-agents/ . Note: the announcement’s claims are as stated by Microsoft in its own release and have not been independently verified.

Leave a Reply

Your email address will not be published. Required fields are marked *