Mercor

AI Safety Experts — English & Portuguese (global)

$29–45/hr

163 people hired

RemoteGlobalResearch3-5 YearsContractPosted Aug 18, 2026

About this role

You will red team conversational AI models and agents to uncover vulnerabilities, misuse cases, bias, and unsafe behaviors. | The role involves adversarial testing through jailbreaks, prompt injections, multi-turn manipulation, and socio-technical probing. | You will generate structured human data and reproducible attack cases that help strengthen AI safety systems. | Native fluency in English and Portuguese (excluding Brazilian Portuguese) is required.

Bilingual AI Safety Expert

Responsibilities

Probe AI models using jailbreaks, prompt injections, misuse scenarios, bias exploitation, and multi-turn adversarial techniques.
Annotate failures, classify vulnerabilities, and identify systemic safety risks using defined taxonomies and benchmarks.
Create reproducible reports, datasets, and attack cases that can be used to improve model safety.
Clearly communicate identified risks and testing outcomes to both technical and non-technical stakeholders.

Qualifications

Native fluency in English and Portuguese (global, excluding Brazilian Portuguese).
Prior experience in AI red teaming, adversarial AI testing, cybersecurity, or socio-technical risk evaluation.
Strong understanding of structured testing frameworks, vulnerability classification, and reproducible evaluation methods.
Excellent written communication, analytical thinking, adaptability, and ability to systematically challenge AI systems.