AppSec Security: The SDLC in the age of agentic
summary
Practical roadmap for securing agentic systems that prioritizes data-lake defence (claims ~50% of breaches stem from data lakes and notes risk when 700 business units share a lake), prescribes splitting lakes, DMZ exports per business unit, strict parameterization and tokenization of emails/credit cards, and encrypting key material so only the right model can decrypt; recommends ingress+egress guard models and kill switches, ownership and lifecycle planning for controls, and lists usable guards like Lama Guard, IBM Granum, GPT-OSS.
AI-written summary · about
chapters (9)
- 0:00Agentic security challenges
- 3:56Prompt injection risks
- 7:05Input sanitization anti-patterns
- 10:59Agentic architecture security
- 15:29RAG pipelines security
- 22:58Securing data lakes
- 30:59Security control lifecycle
- 33:11OWASP resources
- 34:18Q&A
related talks
45mSecurity Architecture, An Adventure?
27mSecurity by design for agentic AI: How CISOs stay in control of autonomous systems
50mAdapting First Order Security Principles to Newer Architectural Trends
43mSecurity By Design
35mGenAI Application Security: Not Just Prompt Injection
30mNavigating the New AI Security Landscape
56mSo Long Secure Coding - Practical Steps for Securing the entire SDLC
51mSecure Applications, By Design