AI Policy Reviewer (Remote)

About the role

The AI Policy Reviewer ensures that AI‑generated and user‑generated content complies with internal governance, regulatory standards, and responsible AI principles. Working remotely, you will act as a quality‑assurance gatekeeper, safeguarding model integrity by identifying safety risks, bias, misinformation, and policy violations.

Key responsibilities

• Review and score AI responses against detailed policy rubrics, assessing safety, truthfulness, fairness, and community‑guideline alignment.
• Identify and flag instances where the model is over‑ or under‑sensitive to policy, handling ambiguous edge cases.
• Analyze data to surface systematic flaws such as bias, hallucinations, or reasoning errors and report them to Product and Engineering teams.
• Collaborate with Policy teams to refine evaluation rubrics and test their teachability to AI models.
• Participate in adversarial testing (red‑team) to uncover jailbreak or unsafe response vulnerabilities before launch.
• Work with Machine Learning Engineers to explain rating decisions and help adjust model behavior.
• Create high‑quality prompt‑response examples for “golden sets” used in model training.

Required profile

• Minimum 3 years of experience in Trust & Safety, Content Policy, Risk Analysis, or Legal/Compliance review.
• Deep knowledge of content moderation areas such as hate speech, harassment, misinformation, and graphic violence.
• Strong analytical ability to deconstruct AI outputs and detect logical flaws, hallucinations, or subtle biases.
• Excellent written communication skills to explain findings to engineers and policy experts.
• Emotional resilience and self‑care strategies for exposure to disturbing content.

Required skills

• Familiarity with large language models (e.g., ChatGPT, Gemini).
• Proficiency with dashboards and spreadsheet tools for data review.
• Experience using specialized content‑review platforms.