Remote AI Safety Adversarial Tester — English & Gujarati

AuraOne; United States; Teletrabalho; USD 22 000 - 30 000; Tempo parcial

Resumo da oferta

AI Trainer Jobs is seeking a remote contractor to perform adversarial evaluation of AI models, crafting prompts and documenting failures with reproduction steps and policy violations. You will score defenses and triage new attack vectors, contributing rigorous patches for modeling teams.

Requirements include red-teaming experience, strong writing, and the ability to work asynchronously for at least 10 hours weekly. Compensation is per hour with program-defined task volume.

Qualificações

• Experience red-teaming AI systems or adversarial ML.
• Strong written communication and clear reporting to patch tickets.
• Familiarity with prompt-injection, jailbreak, and policy-bypass taxonomies.
• Reliable async availability for at least 10 hours per week.

Responsabilidades

• Design adversarial prompts that probe known weakness classes for AI Safety Experts — English & Gujarati assignments.
• Document every successful attack with reproduction steps and the policy clause it violated.
• Score model defenses across single-turn and multi-turn conversations.
• Triage emerging attack vectors and route them to the safety team with severity ratings.

Conhecimentos

Red-teaming AI systems

Security research

Adversarial ML

Strong written communication

Documentation of attempts

Prompt-injection

Policy-grey areas

Multilingual testing

Async availability

Descrição da oferta de emprego

AI Trainer Jobs is seeking a remote contractor to perform adversarial evaluation of AI models, crafting prompts and documenting failures with reproduction steps and policy violations. You will score defenses and triage new attack vectors, contributing rigorous patches for modeling teams.

Requirements include red-teaming experience, strong writing, and the ability to work asynchronously for at least 10 hours weekly. Compensation is per hour with program-defined task volume.