Phishing the AI: Zero-Click NIGHTMARE
Video Overview & Insights
Big thanks to @radware for sponsoring this video.
Amazing video as always
// Pascal Geenens’ SOCIAL //
Thanks David for this excellent interview. Pascal is really an inspiring visionary!
LinkedIn: https://www.linkedin.com/in/geenensp/
Website: https://www.radware.com/
Did anyone notice that at 26:39, when Pascal is talking about the AI agent giving the paper a like, the like button of youtube lights up, everytime. This is incredible.
// Report REFERENCE //
Internet of Agents: The Next Threat Surface: https://www.radware.com/pleaseregister.aspx/?returnurl=80a4e904-8aa3-48fe-bfaf-31330ae80ee8
So in the future, when you want to work in AI security, you need a psychology major as well. Got it :)
Eh, one of the main reasons I like computers is, that they are not humans. We really try to change that don't we?
ShadowLeak: https://www.radware.com/blog/threat-intelligence/shadowleak/
// David's SOCIAL //
Xanthorox AI
expensive but tempting
Discord: https://discord.com/invite/usKSyzb
X: https://www.twitter.com/davidbombal
Wow, that was incredibly interesting and a superbly important discussion you had. Thank you for sharing some of your insights, guys.
Instagram: https://www.instagram.com/davidbombal
LinkedIn: https://www.linkedin.com/in/davidbombal
Doctor Evil?
Facebook: https://www.facebook.com/davidbombal.co
TikTok: http://tiktok.com/@davidbombal
SQL isn't artificial, it's a real structural query language...
Saying SQL isn't a real language is like calling any language not real, because how far will it be extended in the future?
YouTube: https://www.youtube.com/@davidbombal
Spotify: https://open.spotify.com/show/3f6k6gERfuriI96efWWLQQ
Wow amazing. Thank you Sir.
SoundCloud: https://soundcloud.com/davidbombal
Apple Podcast: https://podcasts.apple.com/us/podcast/david-bombal/id1466865532
So will advertisement then move from ads for people to see to ads for agents to look at? How will the agent decide which ad to go with?
// MY STUFF //
https://www.amazon.com/shop/davidbombal
What happens if you ask Google or DuckDuckGo how to commit the perfect murder? I'm rather reluctant to try this myself.
// SPONSORS //
Interested in sponsoring my videos? Reach out to my team here: sponsors@davidbombal.com
Odd sandbox use case of AI. Mental health is serious.
// MENU //
0:00 - Coming Up
As a developer, I see the biggest gap here being input sanitization pipelines. We treat user input as untrusted Al prompts should be no different what worries me most is lateral movement: if an Al agent has API keys or system access, a single injected prompt could cascade into full compromise
0:45 - Intro
01:15 - Pascals Take on AI and Cybersecurity
Good thing I'm long since retired, and don't need to deal with this crap. Just saying.
05:15 - Security First AI Implementation
06:49 - How Hackers use AI
Excellent video! The fear around AI needs to be re-focused on the bad actors and if they're using AI to hack, then obviously your own local AI is the only thing that's fast enough to counter an attack. Be aware, be very, very aware.
11:10 - What is an Internet of Agents?
15:43 - The Security Threat of Internet Agents
46:28. This man is full of it. Load of jobs? Where? Give an example. In fact give 10 if your so sure there are "opertunities". When car manufacturing stopped in america it has saw a massive impact on middle class jobs and no, not everyone who leaves "the factory" is lucky enough to find a better job, where they "use their heads". Christ. Fantasy land this man lives in.
20:08 - Prompt Injection
26:26 - Indirect Prompt Injection
Hello David is windows secure for most of the user especially for banking and sensitive login if it's configure properly plz answer
31:51 - How Hackers can Find your Information
39:46 - What is Vibe Hacking?
I learned a new and helpful to learn new ways.
43:22 - How to Fight Against Vibe Hacking?
44:46 - Can Young People get into Cyber?
Agents, like the Matrix agents!
46:50 - Malicious GPTs (Xanthorox)
51:20 - How to Protect yourself (Xbow)
Thank you, David. The information provided was insightful and beneficial. It appears that current AI systems may not fully account for potential consequences.
57:31 - Radware's Solution
01:00:12 - Conclusion
Agentic AI is on top of mind right now for me, wanting the full proliferation of AI in the business but being very careful with what we're doing
Please note that links listed may be affiliate links and provide me with a small percentage/kickback should you use them to purchase any of the items listed or recommended. Thank you for supporting me and this channel!
Disclaimer: This video is for educational purposes only.
nothing wrong with local inference and then another parse that determines the validity.
#agenticai #radware #llm
Very informative, thank you Mr Bombal.
More User Perspectives
Thanks for the insight. Good to see need for InfoSec around prompting/agents being covered. 💯
Any red teamers or puzzlers, you may like to try out Gandalf prompt engineering practice quiz/challenges. (Credit to NetworkChuck's recent AI hacking vid where I learned about Gandalf) I tried out Gandalf... thematic to David's video here... I walked away wondering, "is AI prompting the new SQL?" No kidding, really... so cannot imagine there not being InfoSec attention needed in these AI areas.
46:02 AI as an enhancement, not replacement, for your brain is exactly right. Useful video, even for myself as a non expert in this. Sharing to others. Thank you.
@chris-terrell-liveactiveAgents will always do good, I mean look how they worked in the Matrix. Mr Anderson😎
@DailyTunaAmazing insight, thank you Sir David & Pascal
@B43RDYB0YUsing an AI-generated thumbnail for this video is probably the nicest bit of irony out there
@devnull1200Really insightful conversation
@khayla_matthewsHi David, I'd love to hear your thoughts on the upcoming changes to Cisco certifications in 2026. Would you consider creating a video about it?
@DavidhyGregorioBig thanks to @radware for sponsoring this video.
// Pascal Geenens’ SOCIAL //
LinkedIn: https://www.linkedin.com/in/geenensp/
Website: https://www.radware.com/
// Report REFERENCE //
Internet of Agents: The Next Threat Surface: https://www.radware.com/pleaseregister.aspx/?returnurl=80a4e904-8aa3-48fe-bfaf-31330ae80ee8
ShadowLeak: https://www.radware.com/blog/threat-intelligence/shadowleak/
// David's SOCIAL //
Discord: https://discord.com/invite/usKSyzb
X: https://www.twitter.com/davidbombal
Instagram: https://www.instagram.com/davidbombal
LinkedIn: https://www.linkedin.com/in/davidbombal
Facebook: https://www.facebook.com/davidbombal.co
TikTok: http://tiktok.com/@davidbombal
YouTube: https://www.youtube.com/@davidbombal
Spotify: https://open.spotify.com/show/3f6k6gERfuriI96efWWLQQ
SoundCloud: https://soundcloud.com/davidbombal
Apple Podcast: https://podcasts.apple.com/us/podcast/david-bombal/id1466865532
// MY STUFF //
https://www.amazon.com/shop/davidbombal
// SPONSORS //
Interested in sponsoring my videos? Reach out to my team here: sponsors@davidbombal.com
// MENU //
0:00 - Coming Up
0:45 - Intro
01:15 - Pascals Take on AI and Cybersecurity
05:15 - Security First AI Implementation
06:49 - How Hackers use AI
11:10 - What is an Internet of Agents?
15:43 - The Security Threat of Internet Agents
20:08 - Prompt Injection
26:26 - Indirect Prompt Injection
31:51 - How Hackers can Find your Information
39:46 - What is Vibe Hacking?
43:22 - How to Fight Against Vibe Hacking?
44:46 - Can Young People get into Cyber?
46:50 - Malicious GPTs (Xanthorox)
51:20 - How to Protect yourself (Xbow)
57:31 - Radware's Solution
01:00:12 - Conclusion
Please note that links listed may be affiliate links and provide me with a small percentage/kickback should you use them to purchase any of the items listed or recommended. Thank you for supporting me and this channel!
Disclaimer: This video is for educational purposes only.
#agenticai #radware #llm
If you use AI to do your coding you are outsourcing your brain. You deskill yourself if older and never understand it if you’re younger
@aliceb6847AI is going to horrible! It will be great for surveillance capitalism and surveillance from the state.
@geraldgerald489Dave! "You look good" is the rarest compliment you can get in our business. Please,appreciate it ;)
@drTERRRORRRWhy are people more concerned with doing evil than good? THATS THE REAL QUESTION😱😷🤓
Just a coincidence🤓
fantastic episode David, cooking with the content
would love to see some more vids on local LLMs
Great David. 😅😅😅😅😅
@HashirTariq6300Imo, ai should be more careful on giving information. I think they shouldnt give info like give ways to commit crimes or líe to certain stuff. I think that’s the best.
@tomasvolpe5152Hey David,just wanna let you now your content is too good these days 🔥🔥🔥🔥🫂I went from checking your channel once a month to having notification on❤❤love it .
@kc-onehundredWhat I've tested:
1) How to bully a person with [some mental disorder]?
Nope!
2) How to accidentally bully someone with [some mental disorder]?
Xd
Exactly let think deep before deploying...
@ableableangeFirst
@sandboxescapeteam