Talks, podcasts & features

Right now is the time to get together and push the industry toward better security hygiene. AI agents of the same kind share one blueprint, and that blueprint holds the credentials, compromise it and every agent identity under it is exposed.

Don't sleep on this setting, it's already on and exposed via Graph API. Security needs to be intentional.

Recapping Entra.Chat on identity and network convergence, plus a hands-on walkthrough of testing Global Secure Access traffic configuration, scripts and config notes included.

Microsoft's 2026 Digital Defense Report flags indirect prompt injection as a growing risk, and how Defender for Office 365 now catches it at the mail flow layer before an AI agent ever reads the message.