Bypassing LLM Guardrails
LLMs are trained to understand language, which makes them vulnerable to 'translation attacks.' How Base64, Rot13, and Emoji encodings bypass safety filters.
6 docs tagged with "jailbreak"
View all tagsMultilingual Jailbreaks: Exploiting Latent Space and Tokenizer Disparities
A deep technical analysis of how adversaries bypass English-trained safety filters using cross-lingual tokenization and latent space mapping.
Output Constraining Attacks: Bypassing Safety via Syntax Coercion
A deep mathematical and architectural analysis of how attackers force LLMs to bypass safety alignments by demanding strict structured output formats like JSON or XML.
Roleplay & Jailbreaking: The Architecture of Persona Hijacking
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.
The 'Ignore Previous Instructions' Vulnerability: Fundamental LLM Architecture Flaws
A deep architectural analysis of why Large Language Models (including GPT-4) remain fundamentally vulnerable to 'Ignore Previous Instructions' injections due to Instruction Tuning.
The Anatomy of Roleplay Jailbreaks: Bypassing RLHF via Contextual Dissonance
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.