9 docs tagged with "red-teaming"

LLMs are trained to understand language, which makes them vulnerable to 'translation attacks.' How Base64, Rot13, and Emoji encodings bypass safety filters.

Polyglot files are valid in multiple formats simultaneously (e.g., GIF + Shell Script). Learn how attackers use them to bypass RAG ingestion filters and achieve RCE.

A deep dive into how adversaries exploit PDF XRef tables and DOM rendering layers to hide prompt injections from humans while guaranteeing LLM execution.

A deep technical analysis of how adversaries bypass English-trained safety filters using cross-lingual tokenization and latent space mapping.

Multimodal RAG systems are vulnerable to adversarial images. Learn how 'Typographic Attacks' and perturbation can trick OCR engines and Vision Transformers.

A deep architectural analysis of persona-based attacks on LLMs. How DAN and Developer Mode exploits manipulate latent space, and how to detect them via structural heuristics.

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.