Japanese Trust & Safety Data Trainer

This is a remote, hourly-paid contract role where you will review AI-generated content and safety decisions, evaluate reasoning quality and step-by-step problem-solving, and provide expert feedback so outputs are accurate, logical, and clearly explained. You will assess solutions for correctness and clarity, spot methodological or conceptual errors, fact-check where needed, and rate/compare multiple responses based on safety and policy alignment. You must be fluent/proficient (near-native or native) in Japanese and able to make nuanced judgments across Japanese and English content. This role is with the fast-growing AI Data Services company SME Careers (a subsidiary of SuperAnnotate), supporting many of the world’s largest AI companies and foundation-model labs. Your annotations on explicit safety tasks will be used to prevent the model from generating unintentional or adversarial, toxic, or unsafe outputs. You may be exposed to content that is sexual, violent, or psychologically disturbing in nature as part of this work to improve the world’s premier AI models.

Key Responsibilities

Label and quality-check safety data across categories such as hate/harassment, sexual content, self-harm, violence, bias, illegal goods/services, malicious activities, malicious code, and deliberate misinformation.
Perform red-teaming and adversarial testing by identifying realistic attack patterns, edge cases, and policy gray areas; document rationales and recommend mitigations to reduce unsafe outcomes.
Apply and localize safety policies consistently across Japanese and English: detect cultural nuance, slang, coded language, and context shifts; escalate uncertainty using documented decision paths.
Show more Show less