8 docs tagged with "prompt-injection"

An analysis of how attackers bypass intent-based NLP filters using Byte-Pair Encoding (BPE) manipulation and Base64 obfuscation to execute hidden payloads.

An analysis of indirect prompt injections in legal documents and how to deploy semantic ML firewalls to protect proprietary RAG knowledge bases.

Advanced architectural strategies for securing Retrieval-Augmented Generation (RAG) pipelines against Indirect Prompt Injection, zero-width Unicode steganography, and SSRF payloads.

A deep architectural analysis of Server-Side Request Forgery (SSRF) in AI applications, targeting the 169.254.169.254 Cloud Metadata Service to steal IAM credentials.

Attackers are hiding prompt injections in zero-width spaces and tabs. Learn how Whitespace Steganography works and why regex is the best tool to catch it.

An analysis of persona-adoption exploits (like the 'Grandma Exploit') that bypass Reinforcement Learning from Human Feedback (RLHF) guardrails, and how to enforce deterministic boundary control.

A comprehensive list of prompt injection techniques for testing RAG systems. From direct overrides to context switching and payload splitting.

An architectural breakdown of Indirect Prompt Injections targeting RAG pipelines, and how poisoned chunks in ChromaDB or Pinecone compromise LLM responses.