Red teaming in the AI era

Module 1: Red Teaming with AI – New Adversarial Capabilities

• The Collapse of Attacker Cost: How to reassess likelihood and frequency in your risk models given AI-driven speed.
• New Classes of Exposure: Measuring readiness against AI-driven social engineering, automated exploitation and adaptive malware.
• Mindset Evolution: Transitioning from point-in-time engagements to continuous, AI-accelerated adversary emulation.

Module 2: Red Teaming of AI — Securing the New Asset Class

• Probabilistic Risk: Understanding why AI systems fail differently than traditional software due to emergent behaviors and expanded blast radius.
• Risk Evaluation: Deep dive into input manipulation, data/model integrity and agentic autonomy.
• Frameworks & Scoring: Mapping AI risks to MITRE ATLAS, OWASP LLM Top 10 and NIST AI RMF.

 

Guest Spotlight: Deep Dives from the Field

Guest Speaker: Eli Shlomo

About
Offensive security is not just a discipline but a way of thinking. Once it becomes part of how you operate, systems, risks and behaviors become much clearer.
Session description
A practical deep dive into EntraReaper demonstrates how Microsoft 365 environments can be compromised through identity abuse

Guest Speaker: Yossi Sassi

Session description
(In)Secure Remote Operations: What Red Teams should use & what to avoid

Guest Speaker: Oren Elimelech

Session description
From Bot to Breach: How HackerBot-Claw Redefined Developer Risk in AI-Driven Supply Chain.

Guest Speaker: Oded Vanunu

Session description
“When AI Agents Become Backdoors: The New Era of Client-Side Attacks” Sharing research on critical vulnerabilities in AI tools & platforms such: Cursor IDE, Anthropic Claude Code, OpenAI Codex CLI, Google Gemini CLI and more affecting millions of users. The session will show how flaws transform trusted AI tools into persistent backdoors, demanding a rethinking of trust models in AI-assisted development.

Key Takeaways

A clear framework for identifying and prioritizing AI-era risks—both offensive and defensive.
Practical approaches to measuring risks that traditional metrics miss.
A modern red team mindset built for probabilistic systems and AI-accelerated adversaries.