Agentic security doesn't need a whole new definition – you just need to reframe what you already know
THE SO WHAT
If AI agents are just new identities with overpowered permissions, the work is extending IAM, monitoring, and least-privilege—not inventing a new security discipline. This week, force every agent project to answer: what identity does it assume, what can it touch, and who’s on the hook when it misbehaves.
READ THE SOURCE
MORE FROM THE WIRE
Applied AITarget SVP says its real AI moat isn't the models — it's everything built around them
Enterprise advantage is consolidating around orchestration, data plumbing, and change management — not who has the flashiest model. If you’re still treating “AI” as a feature, you’re behind the operators who are treating it as an operating system for their business processes.
Applied AIIt’s hard to feel sorry for AI companies when China is giving them a taste of their own medicine
Regulatory and competitive pushback on AI platforms — especially from China — is hardening into a two-way rights and leverage contest. Multinational operators need a clear view of how their data, IP, and models are treated on both sides of that divide before committing distribution or infra.
Applied AIWe’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
Lyria 3.5 folding directly into Flow Music means high-quality, controllable AI music is becoming a native surface, not a side tool. Labels, production houses, and rights holders need explicit policies on how staff and artists can use these systems before they show up in every demo and temp track.
Applied AIClaude and ChatGPT’s latest models are nitpicky and burning tokens. Here’s the fix
More capable models that over-index on nitpicking and verbosity are quietly inflating token bills and latency. Operators should be tuning system prompts and response constraints as aggressively as they tune model choice — verbosity control is now a cost lever.