Uploaded June 2026 | Updated November 2026, 1 day ago
What happens when an unpredictable, confident AI agent gets write access to your production database?
In this webinar clip, Miki dives into the critical security risks of using LLMs for coding and data pipelines. While LLMs are powerful and fast, they lack a native "security mindset" and can confidently make catastrophic mistakes—like deleting production data without a WHERE clause.
We explore real-world examples of vulnerable AI-generated code, why enforcing the principle of least privilege is essential, and how to protect your applications from common AI pitfalls.
Topics covered in this video:
● Analyzing AI-Generated Code: Why basic networking operations generated by LLMs might lead to memory crashes if response bodies aren't size-limited
● The AI Security Mindset: Why you should prompt your LLM to run a security audit on the very code it just generated
● Enforcing Least Privilege: The crucial difference between trusting a developer and trusting their AI agent. Learn why creating "read-only" database users is essential to prevent accidental data deletion
● OWASP Top 10 for LLMs: A look into the OWASP Top 10 risk mitigations for LLMs and GenAI applications, including threats like data poisoning
---
Explore more from Ardan Labs
Online Courses: ardanlabs.com/education
Live Training Events: ardanlabs.com/live-training-events
Technical Blog: ardanlabs.com/blog
Github: github.com/ardanlabs
Connect with Ardan Labs
Website: ardanlabs.com
X: https://x.com/ardanlabs
LinkedIn: linkedin.com/company/ardanlabs
#llmsecurity #aiagents #owasp #genai #cybersecurity #securecoding #leastprivilege #datasecurity
What happens when an unpredictable, confident AI agent gets write access to your production database?
In this webinar clip, Miki dives into the critical security risks of using LLMs for coding and data pipelines. While LLMs are powerful and fast, they lack a native "security mindset" and can confidently make catastrophic mistakes—like deleting production data without a WHERE clause.
We explore real-world examples of vulnerable AI-generated code, why enforcing the principle of least privilege is essential, and how to protect your applications from common AI pitfalls.
Topics covered in this video:
● Analyzing AI-Generated Code: Why basic networking operations generated by LLMs might lead to memory crashes if response bodies aren't size-limited
● The AI Security Mindset: Why you should prompt your LLM to run a security audit on the very code it just generated
● Enforcing Least Privilege: The crucial difference between trusting a developer and trusting their AI agent. Learn why creating "read-only" database users is essential to prevent accidental data deletion
● OWASP Top 10 for LLMs: A look into the OWASP Top 10 risk mitigations for LLMs and GenAI applications, including threats like data poisoning
---
Explore more from Ardan Labs
Online Courses: ardanlabs.com/education
Live Training Events: ardanlabs.com/live-training-events
Technical Blog: ardanlabs.com/blog
Github: github.com/ardanlabs
Connect with Ardan Labs
Website: ardanlabs.com
X: https://x.com/ardanlabs
LinkedIn: linkedin.com/company/ardanlabs
#llmsecurity #aiagents #owasp #genai #cybersecurity #securecoding #leastprivilege #datasecurity










