As you know, I like to try to write a series of blog posts to explain more complex features or something I am working on. I recently wrote a series here about prompting for Artificial Intelligence:
- AI Red Teaming and Adversarial Testing: Building Safer and More Reliable AI Systems (Part 1)
- AI Red Teaming and Adversarial Testing: Designing Effective AI Red Team Exercises (Part 2)
- AI Red Teaming and Adversarial Testing: Evaluating AI Systems Under Adversarial Conditions (Part 3)
- AI Red Teaming and Adversarial Testing – Practical AI Red Team Prompt Engineering (Part 4)
- AI Red Teaming and Adversarial Testing – Thinking Like an AI Red Team (Part 5)
- AI Red Teaming and Adversarial Testing – Building Enterprise AI Evaluation Frameworks (Part 6)
- Part 1: Understanding AI Jailbreaks
- Part 2: Advanced AI Jailbreak Techniques and Real-World Attack Scenarios
- Part 3: Building AI Systems That Can Defend Themselves
- Testing Instruction Hierarchy and Prompt Override Attacks
- Testing Role-Playing and Context Manipulation
- Testing Linguistic Obfuscation and Prompt Encoding
- Advanced Prompt Injection Techniques
I am also in the middle of writing a series on an AI lab I have been building and trying out; you can read more here:
- AI Agents for Security Reconnaissance [Lab Part 1]
- Inside the Lab [Lab Part 2]
- Designing Safe Tools for Security Agents [Lab Part 3]
- The Reconnaissance Agent [Lab Part 4]
- The Application Security Specialist [Lab Part 5]
- The Analysis Agent [Lab Part 6]
- The Validation Agent [Lab Part 7]
- The Defender Agent [Lab Part 8]
Lastly, my latest blog on my company site, SharePlicity, covers the new Conditional Access Policy Plans for Entra ID.