What security leaders said, week 23

June 1, 2026 to June 7, 2026 · 1 entry from 1 leader

Grouped by topic: a week needs at least three entries before we draw up themes. Every person here is someone our record names as the current holder of a top security seat; role and company come from our record.

AI and agents

1 entry on AI tools, agents and their place in security programs.

June 1, 2026· Podcast appearance, CyberWire Daily
Heather Ceylan, Chief Information Security Officer at Box
On artificial intelligence integration in military command structures
AI joins the chain of command.
Our take
Box's CISO appears on this CyberWire Daily Industry Voices segment. The guest segment covers governed AI and the unstructured data problem, set inside a news roundup that also touches battlefield AI, a Netlogon flaw under active attack, and CIRCIA reporting debates. We think the pairing is apt: most governance programmes for AI agents begin with policy and model approval, yet the exposure sits in the sprawl of documents, messages and files that an agent will happily retrieve. Permissions on that content are usually inherited from years of loose sharing, so an assistant simply surfaces what people were always technically able to see. We would start by inventorying where unstructured content lives and who can reach it, and fix oversharing before switching on any retrieval feature. Classification labels only help if the agent's access layer honours them at query time, so test that directly rather than trusting vendor claims. Watch whether agent identity and audit logging mature quickly enough to let teams reconstruct what an agent read and why.
Listen at CyberWire Daily ↗