EWS: Securing Databases in the Age of AI beyond Compliance and DBAs
Join FS-ISAC for an Expert Webinar
Financial institutions face cybercriminals, nation-state actors, and insider threats that are all drawn to the same target: the databases that hold customer information, transaction records, and proprietary intelligence.
Before AI, databases could only be accessed by business users through applications or by DBAs through querying tools. To find any specific data, a user or a DBA had to know which app or table held it, as well as how to use that app or query that table. That friction was a boundary: even a user or a DBA with excessive access was extremely unlikely to fully use it. AI removes this friction. Copilots, agents, and MCP servers reach data by connecting directly to databases, files, and cloud stores. When they inherit the same over-exposed data that human users sit on, they read, analyze, and act on everything they can access, at machine speed.
To give security leaders a way to secure agentic AI, Forrester introduced AEGIS: Agentic AI Enterprise Guardrails for Information Security. It rests on three core principles:
-
Least agency: giving an AI system only the minimum permissions and capabilities a task requires
-
Continuous assurance: verifying security posture live and in real time rather than through periodic audits
-
Explainable outcomes: the ability to reconstruct any decision to allow or block an agent action, trace it to a cause, and explain it to regulators and executives
What you'll learn:
-
How to implement Forrester's AEGIS framework across your data estate
-
Why the application layer that once protected data is dissolving, and what that means for database security
-
How AI agents inherit over-permissioned access and quietly expand your blast radius
-
The questions every security team should be able to answer about its own data