AI Security Engineer Foundations + Certificate — Micah Silverman, Snyk
Micah Silverman uses that capture the flag exercise to connect AI risks with familiar application security controls.
Proof1 independent outlet
Key points
- A student chatbot refuses to reveal grades, but its refusal exposes that it has access to them.
- The six module course covers OWASP risks, shadow AI, threat modeling, MCP security, secure agentic development and red teaming.
- Direct and indirect prompt injection, poisoned tools and excessive agency all return to the same question: what can this system actually access and execute?
- 2:57 - A hacker's mindset and variable model behavior
Sources (1)
- [1]AI Security Engineer Foundations + Certificate — Micah Silverman, SnykAI Engineer (YouTube) · Oct 11, 01:30 PM
Micah Silverman uses that capture the flag exercise to connect AI risks with familiar application security controls.
A student chatbot refuses to reveal grades, but its refusal exposes that it has access to them.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 10, 2026EviScene: Beyond "Is It AI?" for Agentic Image Factuality Assessment
- Oct 10, 2026LightVAE: Towards Compact and Efficient Video Autoencoders
- Oct 9, 2026Microsoft's Decision-1 model enters the fast-growing AI decision model race
- Oct 9, 2026Show HN: Edi Life OS – self-hosted life dashboard with an MCP server for AI
- Oct 9, 2026OpenMobile-2: Building Versatile Mobile Agents with Scalable Environments and App-Native Tools
- Oct 8, 2026openai/codex rust-v0.162.0: 0.162.0
