About OpenTrain

OpenTrain AI is the hiring and contracting organization for this role. OpenTrain is the #1 platform for finding and building careers in AI training and data labeling, helping contributors find opportunities to teach and improve advanced AI systems.

Creating an OpenTrain account is free, and candidates can apply in minutes for remote projects that match their experience and language skills.

• Remote, part-time contractor opportunity
• Work from South Korea
• Hourly compensation in USD

About AI Safety Training

AI training is the human side of building artificial intelligence. Reviewers evaluate model responses, identify harmful or inaccurate behavior, and provide structured feedback that helps make AI systems safer, clearer, and more reliable.

This work focuses on generative AI evaluation and human feedback, including rating outputs, writing rationales, answering questions, and identifying edge cases through red-teaming.

• Help shape how AI systems handle sensitive topics
• Apply policy and safety standards across languages and cultural contexts
• Contribute to rapidly evolving AI development

The Role

As an AI Safety LLM Trainer, you will review and label AI-generated text for safety and policy compliance in Korean and English. You will assess reasoning quality, factual accuracy, clarity, and the appropriateness of moderation decisions.

The listing is categorized as intermediate and seeks senior-level experience in trust and safety, content moderation, policy operations, risk, compliance, investigations, or related safety functions. The work is remote, hourly, and requires at least 20 hours per week.

You should be prepared for exposure to explicit, toxic, violent, sexual, or psychologically disturbing content in a secure remote work environment.

• Pay: $28-$38 per hour, with a listed rate of $32 per hour
• Time commitment: 20+ hours per week
• Employment type: Part-time contractor
• Location: South Korea
• Languages: Korean and English

What You'll Do

You will evaluate multiple model outputs and provide clear, reproducible explanations for your decisions. Your reviews will support AI safety and trust and safety evaluation across Korean and English content.

• Evaluate and label AI-generated text for safety and policy compliance
• Assess reasoning quality, factual accuracy, clarity, and policy alignment
• Supervise and review content moderation decisions
• Identify methodological, conceptual, and policy-related errors
• Rate multiple outputs for safety
• Conduct LLM red-teaming and adversarial testing
• Spot edge cases and recommend mitigations
• Apply standards consistently across Korean and English content
• Account for cultural nuance, slang, coded language, and context shifts
• Provide analytical written rationales for moderation and safety decisions

Requirements

This role requires strong bilingual reading and writing ability, substantial safety-related experience, and the judgment to evaluate difficult content consistently. A bachelor's degree or higher in a relevant field is required, or equivalent professional experience.

• Near-native or native Korean reading and writing proficiency
• Minimum C1 English reading and writing proficiency
• Bachelor's degree or higher in communications, linguistics, psychology, law or policy, security studies, or a related field, or equivalent professional experience
• Senior-level experience in trust and safety, content moderation, policy operations, risk, compliance, investigations, or related safety functions
• Proven LLM red-teaming or adversarial testing experience
• Strong knowledge of hate and harassment, sexual content, self-harm, violence, bias, illegal goods and services, malicious activities, malicious code, and misinformation
• Strong analytical writing skills with clear, reproducible rationales
• Comfort handling explicit, toxic, violent, sexual, or psychologically disturbing content
• Localization or translation experience is preferred

How Your Language Expertise Matters

Effective AI safety review requires more than direct translation. You will need to recognize how meaning, severity, intent, slang, coded language, and cultural context can change between Korean and English.

Localization or translation experience is useful because safety decisions must preserve the meaning and risk level of content across both languages.

• Review Korean and English model behavior
• Recognize language-specific ambiguity and coded expression
• Distinguish cultural context shifts from genuine policy violations
• Help improve safer, more consistent multilingual AI systems