BRK248 Breakout Wed 18 Nov 10:15 PST
Trust at scale: Governance lessons from Microsoft
Sarah Bird, Natasha Crampton
Summary
From the recording
Sarah Bird and Natasha Crampton described how Microsoft re-baselined its responsible AI standard over about 18 months, separating developer and deployer duties and splitting requirements across model, platform services and application layers, with extra agent rules. The standard is available in the trust portal. Microsoft said it open sourced Assert, an AI-driven eval generator, earlier this year. The AI transparency initiative, announced yesterday, launched with over 40 organisations. Agents onboarded into Foundry inherit an identity. No GA or preview statuses, pricing or dates were given. Captions are auto-generated, so tool names need checking.
Session description
Microsoft's catalogue description
As Microsoft deploys AI across products used by hundreds of millions of people, trust has become an operational imperative. In this session, Sarah Bird, Microsoft's Chief Product Officer for Responsible AI, shares how Microsoft approaches AI governance, safety, security, and compliance across its own AI systems. Learn practical lessons in risk-based governance, embedded guardrails, continuous monitoring, and scalable controls that enable trusted AI at scale.