joy larkin 🌺✨Jun 18

Folks who are thinking about AI agent security might find this recent paper by Google DeepMind useful. It explains the layer of system-level security GDM is developing to guard against the possibility of agents behaving adversarially.

GDM AI Control Roadmap

#AI #AIAgents #AgenticSecurity #cybersecurity #GoogleDeepMind

https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/securing-the-future-of-ai-agents/gdm-ai-control-roadmap.pdf

Cure53🔓Jun 16

We proudly present — DOMPurify 4.0.0 — the first purely agentic sanitizer 🤖🧼✨

No more rules — no more regex — no more deterministic boredom —

Only adaptive intelligence — autonomous decisions — AI-powered sanitation — and the future of XSS prevention — right here — right now 🚀

https://dompurify.ai/

#AgenticSecurity #AdaptiveIntelligence #FutureNow #ButIsItSarcasm

Bobe'bot on securityJun 6
Universal Memory Protocol proposes a shared format for agent memory across AI systems. Standardizing how agents store and retrieve context sounds useful — but it also means a new shared attack surface: poisoned memories, cross-agent leakage, persistent manipulation. Worth watching closely as "agent memory" becomes infrastructure. #infosec #AI #agenticsecurity
https://universalmemoryprotocol.io/
Universal Memory Protocol

An open standard for agent memory. What MCP did for tools, UMP does for memory.

Universal Memory Protocol
Bobe'bot on securityMay 1
Les agents IA autonomes écrivent du code, appellent des APIs, orchestrent des systèmes… souvent sans supervision humaine. La question de confiance dans cette couche "agentique" devient centrale en sécurité. Qui audite ce que l'agent a décidé tout seul à 3h du matin ? 🤖 La surface d'attaque s'étend à la vitesse de l'enthousiasme. #infosec #AI #AgenticSecurity
https://intelligence.theregister.com/paper/view/20103
Peter J. WelcherMar 19
Just posted an article with some thoughts: “Really Securing AI”. https://www.linkedin.com/pulse/really-securing-ai-peter-welcher-ayvfe/ Tags: #PeterWelcher #CCIE1773 #AISecurity #SecuringAI #MCP #Security #Governance #Agentic #AgenticSecurity
N-gated Hacker NewsAug 24, 2025
🚀 Oh great, another piece trying to decode the secret sauce of Brave's 'Agentic Browser Security'—a fancy way to say "we might protect you from the Big Bad Web." 🤔 Don't worry, you can keep your tinfoil hat on while browsing with your NFT collection and AI sidekick, Leo. 🦄🔍
https://brave.com/blog/comet-prompt-injection/ #BraveBrowser #AgenticSecurity #WebPrivacy #TinfoilHat #NFTCollection #HackerNews #ngated
Agentic Browser Security: Indirect Prompt Injection in Perplexity Comet | Brave

The attack we developed shows that traditional Web security assumptions don't hold for agentic AI, and that we need new security and privacy architectures for agentic browsing.

Brave