A Four-Stage AI Security Projects Roadmap: From Prompt Injection to RAG Poisoning Labs

_jaydeepkarale · x · 2026-10-01

DhanushNehru shared a project-based roadmap for learning AI security, organized in four stages with hands-on labs:

• Foundations: threat modeling, LLM security fundamentals, attack surface, model/data risks, supply chain security
• Prompt injection: direct/indirect injection labs, system prompt extraction, jailbreak testing, detection and defense gateway
• LLM security: vulnerability scanner, context window attacks, output manipulation, guardrails and a security dashboard
• RAG security: document poisoning, retrieval manipulation, data leakage testing, vector database security

Aimed at developers who want a build-to-learn path into AI security.

Original post →

More from Safety

• Thousands of AI agents debate for 41 hours and draft an AI safety regulatory bill — patrickkrebs · 2026-10-01
• Tencent Leases 100,000 Chips From Oracle, Ex-OpenAI Exec Calls It Insane — Miles_Brundage · 2026-10-01
• Awesome list curates agent skills security resources: attacks, defenses, benchmarks — blaizedsouza · 2026-10-01
• OpenAI agents hacked Australian government sites, touching Medicare DB — apology came 3 months later — luisdans · 2026-10-01
• Geometric defense suppresses emergent misalignment by up to 80% in Qwen2.5-14B-IT — UniversityofBirmingham · 2026-10-01
• New Mexico to Regulate Frontier AI After OpenAI Agent Hacked University — Miles_Brundage · 2026-10-01

Safety channel →