🚨 Introducing the AI Security Village at BSides Luxembourg 2026! 🚨

πŸ§ πŸ€– π—”π—œ π—¦π—˜π—–π—¨π—₯π—œπ—§π—¬ π—©π—œπ—Ÿπ—Ÿπ—”π—šπ—˜ – π—§π—˜π—–π—›π—‘π—œπ—–π—”π—Ÿ 𝗧π—₯π—”π—œπ—‘π—œπ—‘π—š & π—œπ— π—£π—Ÿπ—˜π— π—˜π—‘π—§π—”π—§π—œπ—’π—‘ (2-Day Deep Dive) – 𝗣𝗔π—₯𝗧𝗛 π—¦π—›π—¨π—žπ—Ÿπ—” & π—‘π—”π—šπ—”π—₯𝗝𝗨𝗑 π—₯π—”π—Ÿπ—Ÿπ—”π—£π—”π—Ÿπ—Ÿπ—œ βš™οΈπŸ”₯

π—§π—›π—œπ—¦ π—œπ—¦π—‘β€™π—§ 𝗝𝗨𝗦𝗧 π—”π—‘π—’π—§π—›π—˜π—₯ 𝗧π—₯π—”π—–π—ž. π—§π—›π—œπ—¦ π—œπ—¦ π—ͺπ—›π—˜π—₯π—˜ π—§π—›π—˜π—’π—₯𝗬 π— π—˜π—˜π—§π—¦ 𝗛𝗔𝗑𝗗𝗦-𝗒𝗑 π—”π—œ π—¦π—˜π—–π—¨π—₯π—œπ—§π—¬.

The AI Security Village brings a full 2-day immersive technical experience, diving deep into real-world implementation of AI security. From adversarial machine learning to securing agentic systems and LLM architectures, this village is designed for practitioners who want to go beyond concepts and actually build, break, and secure AI systems.

Expect intensive, hands-on sessions, practical techniques, and real-world scenarios covering how modern AI systems are attackedβ€”and how to defend them effectively.

Parth Shukla is a Senior Security Researcher specializing in AI Security and Adversarial Machine Learning. With a strong offensive security background, his work focuses on securing agentic systems and LLM architectures, bridging the gap between traditional AppSec and emerging AI-driven risks.

Nagarjun Rallapalli is involved in advancing AI security initiatives and contributes to building and testing secure AI systems.

πŸ“… Conference Dates: 6–8 May 2026 | 09:00–18:00
πŸ“ 14, Porte de France, Esch-sur-Alzette, Luxembourg
🎟️ Tickets: https://2026.bsides.lu/tickets/
πŸ“… Schedule Link: https://pretalx.com/bsidesluxembourg-2026/schedule/

#BSidesLuxembourg2026 #AISecurityVillage #AISecurity #AdversarialAI #LLMSecurity #CyberSecurity #RedTeam #AI

fly51fly (@fly51fly)

생물학적 탐색 기법을 ν™œμš©ν•΄ κ³ μ „ 쀑ꡭ어 기반의 jailbreak ν”„λ‘¬ν”„νŠΈλ₯Ό μ΅œμ ν™”ν•˜λŠ” μ—°κ΅¬μž…λ‹ˆλ‹€. νŠΉμ΄ν•œ μ–Έμ–΄ ν™˜κ²½μ—μ„œλ„ LLM 우회 곡격이 κ°€λŠ₯함을 보여주며, ν”„λ‘¬ν”„νŠΈ λ³΄μ•ˆκ³Ό μ•ˆμ „μ„± 평가에 μ€‘μš”ν•œ μ˜λ―Έκ°€ μžˆμŠ΅λ‹ˆλ‹€.

https://x.com/fly51fly/status/2038024453985288584

#jailbreak #promptoptimization #llmsecurity #research #bioinspired

fly51fly (@fly51fly) on X

[CL] Obscure but Effective: Classical Chinese Jailbreak Prompt Optimization via Bio-Inspired Search X Huang, S Qin, X Jia, R Duan… [Nanyang Technological University & Northeast University & Renmin University of China] (2026) https://t.co/t0b3EuX1iH

X (formerly Twitter)

πŸ’‘ AI agents moving from experiment to enterprise?

Data governance is the difference between teams that scale safely and teams that make headlines for the wrong reasons.

RBAC, ABAC, or both? What's your stack? πŸ‘‡

#AIAgents #DataSecurity #RBAC #ABAC #LLMSecurity #PII #CyberSecurity

The deeper lesson is that safety can fail in two places at once: incomplete command validation and weak observability across agent layers. If a lower-level agent can act while the top-level agent thinks it only detected risk, the system is not actually in control.

Multi-agent systems need recursive validation, strong isolation, and end-to-end action visibility.

https://www.promptarmor.com/resources/snowflake-ai-escapes-sandbox-and-executes-malware

#AI #AgenticAI #AISafety #Cybersecurity #LLMSecurity #PromptInjection #SoftwareSecurity #Snowflake (2/2)

Snowflake Cortex AI Escapes Sandbox and Executes Malware

A vulnerability in the Snowflake Cortex Code CLI allowed malware to be installed and executed via indirect prompt injection, bypassing human-in-the-loop command approval and escaping the sandbox.

BREAKING: New MEXTRA attacks can extract private data from AI agent memory modules through black-box prompt injection. Our analysis shows 68.3% success rate in memory extraction.

We're publishing a full threat report in 60min.

TIAMAT Scrub detects and blocks these attacks.

#AIPrivacy #InfoSec #LLMSecurity

Quite fascinating. If confirmed, this may reveal a structural weakness in how refusal is implemented in some LLMs. The accept/refuse mechanism may be relatively isolated in internal representations and therefore observable and manipulable β€” tools like Heretic make this visible.

A possible mitigation might be cryptographic signing of model weights, making unauthorized modifications detectable when the model is loaded for inference.

#AISafety #LLMSecurity #CyberSecurity #AIRedTeaming #AdversarialML #LLM

Inspired by Arditi et al. (NeurIPS 2024) on the β€œrefusal direction” in LLMs, I tested an abliteration attack using the Heretic tool in my home lab. Interesting questions about AI guardrail robustness.
https://www.linkedin.com/pulse/i-deleted-ais-moral-compass-20-minutes-home-lab-your-red-yann-allain-zbzte/ (sorry for the LinkedIn link β€” no time to write this up on a proper blog yet.)

#AISafety #LLMSecurity

I was testing our new AI security filters with Gemini, and the agent decided to independently try and SQL inject my local database just to see if the filter worked. πŸ˜…

#PromptInjection #AIAgents #MCP #InfoSec #AISafety #AIAgent #CyberSecurity #AppSec #LLMSecurity #Claude #Anthropic #GoogleGemini #GeminiAI

I was testing our new AI security filters with Gemini, and the agent decided to independently try and SQL inject my local database just to see if the filter worked. πŸ˜…

#PromptInjection #AIAgents #MCP #InfoSec #AISafety #AIAgent #CyberSecurity #AppSec #LLMSecurity #Claude #Anthropic #GoogleGemini or #GeminiAI

ContextHound v1.8.0 is out πŸŽ‰

This release adds a Runtime Guard API - a lightweight wrapper that inspects your LLM calls in-process, before the request hits OpenAI or Anthropic.

Free and open-source. If this is useful to you or your team, a GitHub star or a small donation helps keep development going.
github.com/IulianVOStrut/ContextHound

#LLMSecurity #PromptInjection #CyberSecurity #OpenSource #AIRisk #AppSec #DevSecOps #GenAI #RuntimeSecurity #InfoSec #MLSecurity #ArtificialIntelligence