AI Red Teamer — Bilingual English & Swedish Safety Specialist

About the role

We're building a global red-teaming initiative to make AI systems safer, more reliable, and less prone to harmful behavior. We're looking for bilingual English & Swedish speakers to join a specialized team that stress-tests AI models by crafting adversarial inputs, surfacing hidden vulnerabilities, and generating high-quality red-team data used to train safer, more robust models.

This is meaningful, high-impact work at the intersection of language, ethics, and machine learning — ideal for detail-oriented thinkers who care about the responsible development of AI.

What you'll do

• Design and execute adversarial prompts intended to surface bias, misinformation, harmful outputs, or unsafe model behaviors
• Review and evaluate AI-generated responses in both English and Swedish for safety, accuracy, and tone
• Document vulnerabilities, edge cases, and failure patterns with clear, structured written feedback
• Collaborate with a distributed team of safety reviewers to refine testing methodologies
• Apply nuanced judgment when assessing sensitive topics such as bias, misinformation, and potentially harmful behaviors
• Maintain strict confidentiality and adhere to ethical guidelines when handling sensitive content

Requirements

• Native or near-native fluency in both English and Swedish (written and spoken)
• Strong critical thinking and analytical skills, with attention to nuance and context
• Comfort engaging with sensitive or controversial content in a professional, objective manner
• Excellent written communication skills for documenting findings
• Reliable internet access and ability to work independently in a remote setting
• Prior experience with AI evaluation, content moderation, linguistics, or trust & safety work is a plus

Compensation

Hourly rate: $48–$62/hour, based on experience and evaluation performance. This is a flexible, remote engagement suited to professionals seeking meaningful part-time or project-based work.