Listen to this Post
I’m looking forโฆ
Home
1
My Network
Jobs
Messaging
25
Notifications
Me
For Business
Hire with AI
Tony Moukbel
Tony Moukbel
Multi-Talented Innovator | 13 Innovations & 4 Patents | Cyber Security Expert | IT & Ai Engineering | 58 Certifications in Cybersecurity, Forensics, Programming & Electronics Dev.
Lebanon
UNDERCODE TESTING
Profile viewers
59
Post impressions
8
Feed post
View company: Secure Debug
Secure Debug
5h โข
๐จ AI agents need security controls.
Not just better prompts.
As agents start using tools, APIs, memory, data, and automation, the attack surface becomes much bigger.
A secure AI agent architecture should include:
๐น Agent identity & access
๐น Just-in-time tool access
๐น Prompt injection defense
๐น Output & data protection
๐น Risk-based action control
๐น Human-in-the-loop approval
๐น Sandboxed execution
๐น Secure memory management
๐น Cross-agent isolation
๐น Behavioral monitoring
๐น Continuous red teaming
๐น Supply chain security
๐ก My biggest takeaway:
AI agent security is not one layer.
It is a stack.
You need to control:
โข who the agent is
โข what tools it can call
โข what data it can access
โข what actions it can perform
โข how outputs are validated
โข how behavior is monitored
โข when humans must approve
๐จ The biggest risk is not only that an agent gives a wrong answer.
The bigger risk is that it takes the wrong action with real permissions.
That is why agent security must be built around least privilege, isolation, monitoring, and human oversight.
AI agents can automate work.
But without controls, they can also automate risk.
๐ฌ Which control do you think matters most for AI agents?
Prompt injection defense, tool access, sandboxing, monitoring, or human approval?
AISecurity AgenticAI AIGovernance LLMSecurity CyberSecurity AIAgents RiskManagement ZeroTrust DataSecurity PromptInjection DevSecOps InfoSec
View image
33 reactions33
1 comment1 comment
โข
3 reposts3 reposts
Like
Comment
Repost
Send
View Vikramaditya Karanwalโs profile
Vikramaditya Karanwal Verified Profile 2ndVikramaditya Karanwal โข 2nd
๐ก๏ธAI Security | Securing LLMs, RAG & Agentic AI Systems | DevSecOps | Product, Application, Container & Infrastructure Security | Governance, Risk, Compliance & Audit | Shift-Left Security | Learning Spanish & German
22m
Great Share
About
Accessibility
Help Center
Ad Choices
Advertising
Get the LinkedIn app
More
LinkedIn Corporation ยฉ 2026
Tony Moukbel
MessagingYou are on the messaging overlay. Press enter to open the list of conversations.
Compose message
You are on the messaging overlay. Press enter to open the list of conversations.
Page inboxes
Click to see affiliated inboxes
โถ๏ธ Related Video (98% Match):
๐ฏLetโs Practice For Free:
๐ Live Courses & Certifications:
[Join Undercode Academy for Verified Certifications](https://undercode.co.uk/certifications/)
๐ Request a Custom Project:
Secure, high-velocity infrastructure and disruptive technological engineering. Contact our engineering team for high-tier development and proprietary systems:
[[email protected]](mailto:[email protected])
๐ Smart Architecture | ๐ก๏ธ Secure by Design | โญ Trusted by Thousands
IT/Security Reporter URL:
Reported By: [Aisecurity Agenticai](https://www.linkedin.com/posts/aisecurity-agenticai-aigovernance-share-7468244504815337472-1bnQ/) – Hackers Feeds
Extra Hub: Undercode MoN
Basic Verification: Pass โ
๐JOIN OUR CYBER WORLD [ CVE News โข HackMonitor โข UndercodeNews ]
[๐ฌ Whatsapp](https://undercode.help/whatsapp) | [๐ฌ Telegram](https://t.me/UndercodeCommunity)
๐ข Follow UndercodeTesting & Stay Tuned:
[๐ formerly Twitter ๐ฆ](https://x.com/undercodeupdate) | [@ Threads](https://www.threads.net/@undercodetesting) | [๐ Linkedin](https://www.linkedin.com/company/undercodetesting/) | [๐ฆBlueSky](https://bsky.app/profile/undercode.bsky.social)
