Bilingual Arabic–English AI Safety Content Evaluator

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. We help people start and grow careers teaching AI by connecting contributors to projects, hosting profiles, and enabling easy applications—creating accessible pathways into a fast-growing field.

About AI training and safety work

AI training (data labeling/annotation) is the human side of building modern AI. Contributors prepare, review, and evaluate examples that shape how models behave—tasks include reviewing model outputs, red-teaming prompts, and labeling content for safety and accuracy.

This role sits at the intersection of LLM safety and trust operations: your work helps prevent models from producing toxic, adversarial, or otherwise harmful responses, and directly impacts how systems handle sensitive topics.

The role — what you'll do

As an AI Safety Content Evaluator you will review AI-generated responses and create safety-focused evaluation content in both English and Arabic. You will follow written guidelines to assess reasoning quality, annotate outputs for safety concerns, and provide clear feedback and documentation of unsafe model behaviors.

• Evaluate and rate model responses for safety, factuality, and clarity using provided frameworks and rubrics.
• Annotate and label text outputs that include sexual, violent, toxic, self-harm, misinformation, or otherwise sensitive content.
• Identify adversarial prompts and document unsafe or unexpected model behaviors observed during red-teaming.
• Write concise explanations for ratings and decisions so engineering and policy teams can act on findings.
• Work with text-only data types and both evaluation-rating and text-generation labeling tasks using the project’s tooling.

Requirements — skills, experience, and qualifications

You must be comfortable evaluating explicit, toxic, violent, sexual, or psychologically disturbing material as a regular part of the job and apply safety policies consistently when cases are ambiguous.

• Near-native or native proficiency in Arabic reading and writing.
• Minimum C1 proficiency in English reading and writing.
• Bachelor’s degree or higher in Communications, Linguistics, Psychology, Law/Policy, Security Studies, or equivalent professional experience.
• Proven experience in Trust & Safety, content moderation, policy enforcement, risk operations, investigations, or safety evaluation.
• Hands-on LLM red-teaming experience, including identifying adversarial prompts and documenting unsafe model behaviors.
• Strong working knowledge of safety domains: hate and harassment, sexual content, suicide and self-harm, violence, bias, illegal goods/services, malicious activities and code, and misinformation.
• Ability to apply written safety policies consistently and explain decisions clearly in ambiguous cases.
• Comfort with AI tools such as Perplexity, Gemini, ChatGPT, or similar systems.
• Prior experience with AI data training, annotation, or evaluation workflows is preferred.

Who should apply

Apply if you have bilingual Arabic/English language skills, hands-on experience evaluating or red-teaming LLMs, and previous work in trust & safety or content moderation. This role suits mid-level professionals who want flexible, remote contract work that directly shapes model behavior and safety.

• Intermediate experience level; proven judgment handling sensitive material.
• Comfortable annotating and documenting nuanced, adversarial, or harmful content.
• Able to work independently and explain policy-based decisions in writing.

Hours, compensation, and logistics

This is a part-time contractor role requiring 20+ hours per week and is fully remote and worldwide. Pay is per hour; projects list rates between $15 and $40 USD per hour (hourlyRate indicated: $25/hr). You will work with text labeling tasks and use the project’s annotation software (OTHER).

• Employment type: Contractor, Part-time.
• Workload: 20+ hours/week (flexible scheduling dependent on project).
• Pay: $15–$40 USD/hour (hourlyRate: $25 USD).
• Data type: Text; label types: Evaluation rating and text generation.

How the work is managed and next steps

Projects follow strict safety guidelines and documentation standards. You will receive training on the project’s rubrics and access to tooling for submitting evaluations and notes. Expect iterative feedback from project leads to calibrate ratings and documentation.

To apply, create an OpenTrain profile, list your language skills and relevant experience, and submit an application for this project. The platform streamlines onboarding so qualified candidates can begin work quickly.

• You will complete training and calibration tasks before full task access is granted.
• Feedback loops and quality checks are part of the project to ensure consistent evaluations.
• Because of the role’s nature, the ability to maintain personal well-being while reviewing disturbing content is important—please only apply if you are prepared for that exposure.