Joined March 2018
- Tweets 5,392
- Following 208
- Followers 4,809
- Likes 1,573
Photos and videos
May 29
The biggest myth in AI security?
“Internal users are safe.”
In reality, your LLM chat interface might already be a data exfiltration channel.
All it takes is the right prompt.
Here’s what security teams need to understand.
(P.S. Our biggest sale of the year is live — 50% off sitewide with NOEXCUSES50)
1
430
May 26
You don’t get better at security by watching videos.
You get better by breaking things, fixing them, and learning why.
That’s exactly what we built.
🔥 50% off sitewide
Use code: NOEXCUSES50
Stop watching. Start building.
🔗 Link in reply
1
131
May 20
You can’t DevSecOps your way out of bad pipelines.
This bootcamp gets deep into automation, policy, SBOMs, and real security in CI/CD.
No buzzwords. Just hands-on, down-to-the-wire sessions.
🏅 Includes 2 Certification Exam Attempts
🎓 Comes with 1-Year AppSecEngineer Pro Annual Subscription
Use the code ‘BOOTCAMP25’ and get 25% off.
Sign up today — link in the comments.
1
1
107
May 15
Your CVE scanner passing doesn’t mean your software supply chain is secure.
Compromised dependencies, poisoned pipelines, and malicious updates often slip through unnoticed.
@abhaybhargav breaks it down in his latest G2 article👇
learn.g2.com/how-to-identify…
204
May 14
Your AI model may not be hacked.
Its context may just be corrupted.
And that’s what makes context desynchronization dangerous.
No obvious exploit.
Just AI systems confidently making the wrong decisions.
This video breaks down how it happens in modern AI pipelines 👇
(P.S. Sitewide sale is live — 50% off with NOEXCUSES50)
1
98
May 12
Everyone’s debating whether MCP is “dead.”
That’s the wrong conversation.
MCP gives AI agents access to tools.
Skills tell them how to use those tools safely, efficiently, and consistently.
And that’s exactly why MCP security matters.
As AI agents gain deeper access to codebases, databases, and environments, risks like Tool Poisoning, Agent Goal Hijacking, Command Injection, Shadow MCP Servers, and Secret Exposure become very real.
That’s why we launched new hands-on MCP OWASP Top 10 courses—to help security engineers understand how these attacks actually work in modern AI systems.
🎯 Learn the risks. Practice the defenses. Build real AI security skills.
🔥 50% OFF sitewide
Use code: NOEXCUSES50
🔗 Link in comments
2
2
190
May 11
You don’t get better at security by watching videos.
You get better by breaking things, fixing them, and learning why.
That’s exactly what we built.
🔥 50% off sitewide
Use code: NOEXCUSES50
Stop watching. Start building.
🔗 Link in reply
1
1
90
Reading about vulnerabilities is one thing.
Fixing them in real scenarios is another.
AppSecEngineer is built to take you from passive learning → real capability.
If you want to actually apply what you learn, this is where it happens.
Use code: NOEXCUSES50
🔗 Link in reply
1
102
Apr 30
The scariest AI failure mode right now?
Not prompt injection.
Not model compromise.
It’s when the model works perfectly…
…but the context feeding it is wrong.
That’s where things get dangerous.
Watch how this happens in modern AI pipelines.
Learn how to secure AI systems today on AppSecEngineer.
115
Apr 29
Healthcare software is one of the few environments where small security mistakes can scale into systemic risk.
Developers aren’t just writing code — they’re shaping how patient data flows across systems, APIs, and clinical workflows.
Secure coding becomes far more effective when teams move beyond generic guidance and start working with patterns that reflect the realities of healthcare systems.
The goal isn’t perfection.
It’s making the secure way the easiest way to build.
Learn more in the full guide (link in reply).
1
86
Apr 29
“Process this invoice.”
→ Agent issues a refund instead.
@abhaybhargav breaks down goal hijacking—an indirect prompt injection hidden in trusted data.
Watch 👇
165
Apr 28
A chatbot bug in e-commerce is annoying.
A chatbot bug in healthcare can expose protected health information (PHI).
Or surface internal claims data.
Or leak privileged medical records.
AI is powerful.
But in healthcare, it also expands the attack surface.
Watch the full webinar on healthcare security.
Link in the reply.
1
72