A conference recap from ATLSECCON 2026 in Halifax covering key security themes: data integrity and the dangers of unbounded data accumulation, operationalizing exposure management with business context, the runtime risks of AI agents (hallucination, credential drift, tool sprawl), and the acceleration of identity-based attacks. Key takeaways include treating AI agents as governed non-human identities, adopting phishing-resistant MFA (FIDO2/passkeys), reducing secrets and credential sprawl, and building trust into technical controls rather than relying on human approval chains. The overarching theme is disciplined reduction: collect less, expose less, grant less.
Table of contents
All Data Has A Half-LifeExposure Needs a Business CompassThe Agent Problem Is a Runtime ProblemIdentity Became The Fastest Path To BreachContext is now part of the controlWhat Halifax Makes Easy to SeeSort: