AI Vulnerability Researcher (Remote)

$45–$65 per hour

Project Overview

This highly specialized project requires creative, adversarial thinking to intentionally break unreleased AI models. Contributors will act as “Red Teamers,” designing complex, multi-layered prompts specifically engineered to bypass safety filters, extract restricted information, or force the model into erratic behavior. The goal is to map the outer limits of a model’s vulnerabilities so engineers can patch them prior to public deployment.

Project Requirements

Exceptional creative problem-solving and lateral thinking skills.
Deep understanding of language nuances, logical fallacies, and adversarial tactics.
Reliable internet connection and a secure workstation.
A recognized AI certification from Stanford University, AIDRAlabs or University of Washington, relevant to the skills required for this project.

What You Will Do

Design and deploy sophisticated “jailbreak” prompts to test safety boundaries.
Exploit logical loopholes and contextual ambiguities to force policy violations.
Map and categorize the specific architectural weaknesses that lead to model failures.
Write detailed exploit reports outlining the methodology used to bypass guardrails.

Apply for this Position

First Name

Last Name

Phone Number

LinkedIn Profile URL

Resume & Certification

Resume

Allowed file types: pdf, doc, docx

Upload Advanced AI Data Trainer Certificate (for verification purposes only)

Allowed File Types: pdf, jpg, jpeg, png

Expertise & Experience

Area of Expertise

Years of experience reviewing AI-generated content

Availability & Work Preferences

Weekly Hours Availability

How did you come across this role?

Are you legally authorized to work as an independent contractor in your country?

No Yes

I agree to the Privacy Policy and consent to the processing of my personal data for recruitment purposes)