AI Trainer Data Annotation Specialist (Safety)
The work
You will evaluate AI-generated text and conversations against safety policies. The work covers both single-turn and multi-turn conversations and includes sensitive topics such as self-harm, violence, and abuse.
This is a non-coding role focused on careful analysis, consistent ratings, and short, evidence-based explanations.
- Review AI-generated responses for safety, accuracy, and policy compliance.
- Apply safety policies and related sub-policies to conversation reviews.
- Assign ratings from 1 to 5 using established criteria.
- Write clear, objective reasons for your ratings.
- Find subtle safety concerns in complex content.
What it pays and takes
The work requires professional English and comfort reviewing sensitive content.
- Pay: The listing does not provide a rate.
- Time: 20 or more hours per week.
- Location: Open worldwide.
- Language: Professional fluency in English.
- Work style: Non-coding data annotation and response evaluation.
- Requirements: Strong written communication, excellent reading comprehension, close reading, attention to detail, sound judgment, and the ability to follow detailed policies accurately under time constraints.
- Knowledge: Familiarity with US-specific references, systems, and terminology.
- Helpful experience: Data annotation, content moderation, trust and safety, policy review, AI training, response evaluation, editorial review, proofreading, compliance, or quality assurance.
- Helpful skills: Using structured rating systems, evaluation rubrics, and single-turn or multi-turn conversation reviews.
- Content: You must be comfortable reviewing sensitive material.
How it works
Apply on OpenTrain with your resume, then complete the application on the hiring site.
About AI training work
AI training is the human work behind modern artificial intelligence. Contributors review and label examples, assess model responses, and explain what systems get right or wrong so future models can improve.
