AI Safety Red Teamer — Bilingual English & Dutch

About the role

We're building a specialized red team of bilingual language experts to stress-test AI models before they reach the real world. The safest AI systems are the ones that have already been attacked, probed, and challenged by skilled humans who understand nuance, context, and cultural subtlety in more than one language.

As an AI Safety Expert on this project, you'll work directly with AI outputs that touch on sensitive and high-stakes topics — including bias, misinformation, harmful behaviors, and edge-case reasoning failures. Your job is to think like an adversary: craft prompts, evaluate responses, and surface the vulnerabilities that matter most. This is fully remote, text-based work that fits around your schedule.

What you'll do

• Design and execute adversarial prompts intended to surface unsafe, biased, or otherwise problematic AI model outputs
• Review and annotate AI-generated responses in both English and Dutch for safety, accuracy, tone, and cultural appropriateness
• Identify patterns of failure related to misinformation, harmful content, bias, or policy violations
• Document findings clearly so they can be used to retrain and harden AI models
• Collaborate with a distributed team of safety experts to calibrate standards and share emerging risk patterns
• Provide structured feedback and edge-case examples that improve model robustness over time

Requirements

• Native or near-native fluency in both English and Dutch, written and spoken
• Strong critical thinking and analytical skills, with comfort engaging with sensitive or uncomfortable subject matter
• Excellent written communication and attention to detail
• Ability to work independently in a fully remote, asynchronous environment
• Prior exposure to content moderation, linguistics, trust & safety, policy review, or AI/ML evaluation is a strong plus
• Reliable internet connection and availability for consistent part-time or full-time hours

Compensation

This role pays $48–$62 per hour, commensurate with experience and language proficiency. Work is fully remote and open globally to qualified bilingual candidates.