Kicking off our afternoon at @owasp#AppSecIL with a fantastic keynote from @netadror about how hackers are loving the fact that everyone is an AI builder now!
מה קורה כשמזמינים אותי לתת הרצאת keynote בכנס סקיוריטי? האם אני ואנשי הסייבר נסתדר? האם הם יצחקו מהבדיחות שלי והאם אוכל לומר שאני סוף סוף גם בסייבר? 😎
נגלה ביום שלישי הקרוב!
כרטיסים אחרונים פה 👇
https://t.co/IxYp8hcVXi
SESSION SPOTLIGHT
Off Script at Scale: Analyzing Rogue Agents Across 86K Repos and a Year of Production Incidents
Bar Kaduri
A year of production incidents meets analysis of 86,000 repositories in a look at rogue agents operating off script.
https://t.co/GuKIfGgJsn
Every year, we want to professionally record the amazing talks at OWASP AppSec Israel and share them with the community. And every year, the reality is the same — our budget is simply too tight to make it happen.
1/6
Video Sponsorship: $6,500 USD
🔥 Special 15% last-minute discount
There are only 2 opportunities available.
If your company wants to support the AppSec community while leaving a lasting mark on AppSecIL 2026, reach out to me ASAP — we have only 5 days to make this happen!
5/6
SESSION SPOTLIGHT
Sandboxes Are not Enough: Securing AI Agents in a World Full of Secrets
Shiran Bareli
Agents pursue goals, adapt when blocked, and find secrets. Why a sandbox is only a speed bump.
https://t.co/GblpBKbGyG
SESSION SPOTLIGHT
Keynote: Welcome to the Autonomous Attack Surface
Keren Katz, Director of AI Security Research at Zenity and leader of the OWASP Top 10 for Agentic Applications, maps the autonomous attack surface.
https://t.co/U5Ktts7GN1
SESSION SPOTLIGHT
Never Say Never: Owning ChatGPT's Secure Sandbox
Simcha Kosman
Security research tests ChatGPT's sandbox assumptions with a cross-tenant data-exfiltration attack chain.
https://t.co/gW8LBfxN2Z
SESSION SPOTLIGHT
LLMs as Adaptive Query Planners for SAST
Oren Ish-Shalom
Use an LLM to incrementally plan semantic SAST queries and trace data flows—without forcing a whole repository into context.
https://t.co/Qy2lG9gmLv