Security projectby Benchaphorn (Irene) Cho

Cloud Security
Data Lake
with anAI Triage Assistant

I build a test cloud setup with Terraform, run safe, simulated attacks against it, and send all the security logs into a data pipeline I built. Detection rules written in SQL flag the attacks, and an AI assistant writes a short summary of each alert. Then I attack the AI assistant too, and add defenses so it can't be tricked.

pipeline
  1. 01 Terraform
  2. 02 Simulated attacks
  3. 03 Data pipeline
  4. 04 SQL detections
  5. 05 AI triage

Section 01: Section 01

How it works

  1. 01

    Build

    A test cloud setup is built with Terraform.

  2. 02

    Attack

    Safe, simulated attacks are run against it.

  3. 03

    Collect

    All the security logs flow into a data pipeline I built.

  4. 04

    Detect

    Detection rules written in SQL flag the attacks.

  5. 05

    Triage

    An AI assistant writes a short summary of each alert.

Each AI alert summary covers three things

Triage output

  • AWhat happened
  • BHow serious it is
  • CWhat to do next

Section 02: Section 02

Then I attack the AI assistant

The assistant reads the logs, so the logs become a way to attack it.

Attack

Hidden prompt injection

Prompt-injection text is hidden inside the logs to try to trick the assistant.

vs
Defense

Harder to trick

Defenses are added so the assistant is harder to trick.

Section 03: Section 03

Why it matters

Problem one

Security teams drown in cloud logs.

Problem two

AI tools that help them can be turned against them.

This project works on both problems.

Section 04: Section 04

Where to see it

The project repository is on GitHub. The write-up is coming soon.