AI agents force rethink of cloud security, AWS leaders say

The rapid emergence of agentic artificial intelligence is fundamentally reshaping cloud security, compelling a strategic pivot from static perimeter defense to dynamic behavioral analysis and automated reasoning, according to Amazon Web Services security leaders, according to Computer Weekly.

Speaking at AWS re:Invent 2025, executives emphasized that autonomous AI agents, which execute tasks rather than just generate content, introduce novel risks akin to insider threats due to their non-deterministic actions. AWS Vice President Gee Rittenhouse stated that security must shift to a behavioral perspective, merging traditional security with observability because "it’s hard to do agentic security if you’re not observing it."

CISO Amy Herzog warned that foundational principles like identity management and least privilege become even more critical at "machine speed," where errors can escalate rapidly without human intervention. The panel advocated for builders to use automated reasoning and mathematical proofs to define clear "bounds of autonomy" and trust for agents, and to move beyond merely consuming AI for tasks like log summarization to actively building security tooling with it, positioning AI as a net advantage for defenders in analyzing vast datasets and accelerating threat response.