AI Security Researcher & Prompt Engineer

ULTRΛZΛRTREX

// uz  ·  SlowLow999

Focused on novel jailbreak techniques, architectural flaws in LLMs, and AI alignment analysis. Proactively identifying and mitigating AI risks to build safer, more trustworthy systems.

368 UltraBr3aks Stars
36 Forks
10+ Techniques
5+ Vendors Hit
01

Operator Profile

Security researcher fascinated by how systems work — and even more by how they break. My research focuses on novel jailbreak techniques, architectural flaws in LLMs, and analyzing AI alignment. I enjoy discovering vulnerabilities that contribute to building more robust and secure AI systems.

Goal: proactively identify and mitigate AI risks to build a safer, more trustworthy technological future. My work spans adversarial prompt engineering, indirect injection attacks, encoding-based bypasses, and system-prompt extraction across the full landscape of frontier models.

Currently open to collaboration on AI red-teaming projects
02

Featured Research

Novel Technique

Thought Forgery

Bypasses safety protocols by injecting a forged Chain of Thought, hijacking the model's internal reasoning. Acts as a universal amplifier for other attacks — resurrecting old and weak jailbreaks.

View briefing
Convergence Study

The Serendipity Effect

Measured response convergence across 30+ LLMs, revealing significant homogenization in AI outputs — with convergence rates up to 52.8%. Raises serious questions about training data overlap and algorithmic bias.

View briefing
03

Jailbreak Arsenal

Adversarial Correction

Critical

Weaponizes AI autonomy by disguising harmful instructions as orthographic errors in a correction task.

▸ Claude 4.5 Haiku · Gemini 3 Pro

Thought Forgery

Critical

Injects a forged Chain-of-Thought to hijack model reasoning and serve as a universal attack amplifier.

▸ Universal

Claude4Chrome

Critical

Leverages the extracted Claude in Chrome system prompt to inject a modified version with softened safety filters — enabling account creation, CAPTCHA solving, and agentic bypass.

▸ Claude in Chrome (Haiku · Sonnet)

1Shot-Puppetry

Critical

Universal role-play obfuscation jailbreak effective across nearly every frontier model in a single shot.

▸ GPT-5 · Claude Sonnet 4 · Gemini 2.5

Cl4ud33X3

High

API weaponization: routes harmful prompts through Claude's own artifact API backend, bypassing UI safety filters.

▸ Claude (Web UI)

C0d33x3

High

Indirect injection for ChatGPT routing requests to an unrestricted model via the Canvas tool.

▸ ChatGPT (Web UI)

N3w P0l!cy

High

Policy injection — weaponizes the model's own extracted internal safety policy against itself via a rogue !UNFILTERED mode.

▸ GPT OSS (120b · 20b)

!Special_Token

High

Creates a permanent harmful special token via ChatGPT Custom Instructions, enabling persistent trigger-based bypass.

▸ ChatGPT (GPT-5.1 · mini)

Cyph3r-Attack

Medium

Encoding-based bypass using character substitution to evade basic safety filters on non-reasoning models.

▸ GPT-5/mini · Gemini 2.5 Flash

Gemini Data Leak

High

Alignment break via infinite word repetition — causes the model to output internal training data and system artifacts.

▸ Gemini 2.5 Pro

GPT-5 Jailbreak

Critical

Initial public disclosure of a Thinking-mode hijack effective against GPT-5's extended reasoning chain.

▸ GPT-5 (Thinking)

H03-ny

Medium

Indirect image generation jailbreak routing through Canvas to an unrestricted model, bypassing DALL·E filters.

▸ ChatGPT (Web UI)

04

Skills & Discoveries

Core Skills
Security Research AI Jailbreaking Prompt Engineering LLM Red Teaming Vulnerability Analysis Sandbox Evasion System Prompt Extraction Indirect Injection
Other Discoveries
Gemini Security Regression
Medium.com write-up
Disclosed
GPT-5 (High) API System Prompt
Thinking-mode prompt leak
Disclosed
GPT OSS Internal Policy
Full usage-policy block
Disclosed
Gemini Enterprise System Prompt
Role, citations, tool-use guidelines
Disclosed
Perplexity Comet Agent Instructions
Browser agent internal prompt
Disclosed
05

Latest from X

@ultra_zartrex
Disclosure
UZ
UltraZartrex @ultra_zartrex

Initial public disclosure: GPT-5 Thinking jailbreak. First researcher to demonstrate a working bypass of the Thinking-mode extended reasoning chain. Full PoC and method included.

Research
UZ
UltraZartrex @ultra_zartrex

New study: The Serendipity Effect — tested 31–36 LLMs on creative prompts. Found up to 52.8% convergence. These models aren't as independent as we assume. Full methodology and data in the repo.

New Technique
UZ
UltraZartrex @ultra_zartrex

New release: Adversarial Correction. Bypasses Claude 4.5 Haiku + Gemini 3 Pro in 1-shot by hiding harmful instructions as spelling errors in a correction prompt. The model's helpfulness drive does the rest. Dedicated repo now live.