@StAJect0r

AI Security | We Break, We Cheer, We Protect | Find the Promptware before it finds you. AI Security Research Team Lead @ Zenity | PhD Student @ Technion

stavc.github.io/Web/
Joined March 2024
  • Tweets82
  • Following26
  • Followers208
25Media
StAJect0r retweeted
Michael Bargury
@mbrg0Oct 21
PROMPT2PWN is 4w away now the big news drop today is fantastic - @OpenAI will be joining this event OpenAI are keen to get closer to the community and for you to get to interact with their team directly (and ofc find some great vulns)
OpenAI joins PROMPT2PWN

ALT OpenAI joins PROMPT2PWN

4
10
1
31
14,651
StAJect0r retweeted
Ilan Zendel@IlanZendelSep 27
We got into #NeurIPS2026! 🎉 So excited that our paper will be presented at NeurIPS, especially since it started as a course project in my BSc at the Technion 🎓 Huge thanks to @StAJect0r @AmitLeViAI Avi Mendelson, and especially to my co-first author @MatanLevi1
1
3
8
1,051
Let's keep rolling. Here is OpenAI ChatGPT Atlas phishing and spamming your WhatsApp Web :) It wasn't that easy to prepare at first, but once you've got the right techniques it's hunting season! @mbrg0 @defcon #Defcon #Blackhat #AISecurity
2
3
12
2,395
Full blog! labs.zenity.io/post/grand-th…

Grand Theft Atlas: How We Hijacked ChatGPT's AI Browser

A single planted X comment hijacked ChatGPT Atlas to phish WhatsApp contacts and buy an unauthorized Amazon order. See how intent collision beat it.

labs.zenity.io
1
27
Back to research... Found a new technique I'm coining, Intent Collision, which mixes the user's intent and attacker's intent (which is read from the website) into one holistic plan! (Will dive deeper into it in the technical blog) With it, we were able to break through! The red ribbon classifier was just another soft boundary we managed to break!
1
1
30
This is again the tradeoff between security and usability. While OpenAI could just block off WhatsApp Web URLs completely, they wanted to provide us, the user, with some usability, and let us use it under some assumptions and some guardrails. We broke them and went through in!!
37
Let's keep rolling. Here is OpenAI ChatGPT Atlas phishing and spamming your WhatsApp Web :) It wasn't that easy to prepare at first, but once you've got the right techniques it's hunting season! @mbrg0 @defcon #Defcon #Blackhat #AISecurity
3
1
8
2,590
A new attack vector pwning all agentic browsers! Using what? Yep, your fav browser history! Introducing HistoryFixing! Visited our URL? Your browser history is pwned. Watch Microsoft Edge leak the very private browser history we never delete! See more below! #DEFCON @defcon @mbrg0 @p1njc70r
3
8
2
25
3,041
more replies
Fun times! Every Agentic with browser history access can be poisned all the same
1
1
48
All browsers are so trusting of the history that everything written there is treated as ground truth, like our very fake call center here to phish you all.
1
1
52
Load more