AI Safety Experts — English & Norwegian

This position is listed on behalf of a partner company, who manages all applications and next steps. Our partner is looking for an AI Safety Expert — English & Norwegian based in Canada. This is a fully remote opportunity to contribute directly to the development of safer, more reliable AI systems. You will join a red-team-focused project designed to identify vulnerabilities that automated evaluations may overlook. The role combines adversarial thinking, structured testing, and high-quality human data generation. You will challenge conversational AI models through jailbreaks, prompt injection, misuse scenarios, and multi-turn manipulation. Your findings will help strengthen AI safety benchmarks and improve the robustness of systems used in real-world environments. The work is text-based, flexible, and conducted across evolving projects and customer needs. Sensitive topics may arise, but participation in higher-sensitivity projects is optional and supported by clear guidelines and wellness resources. Red-team conversational AI models and agents by testing jailbreaks, prompt injections, misuse cases, bias exploitation, and multi-turn manipulation scenarios. Generate high-quality human evaluation data by annotating model failures, classifying vulnerabilities, and identifying systemic safety risks. Apply established taxonomies, benchmarks, testing frameworks, and playbooks to ensure evaluations are consistent and reproducible. Create structured attack cases, datasets, and reports that clearly document vulnerabilities and provide actionable insights for improving AI systems. Identify weaknesses that automated testing may miss and develop creative adversarial approaches to uncover unexpected failure modes. Communicate technical and socio-technical risks clearly to both technical and non-technical stakeholders. Adapt quickly across different projects, AI systems, evaluation methodologies, and customer requirements. Contribute to expanding evaluation coverage and reducing the likelihood of unexpected AI safety issues reaching production environments. Requirements: Native-level fluency in both English and Norwegian , with excellent written communication skills in both languages. Prior experience in red teaming, AI adversarial testing, cybersecurity, socio-technical probing, conversational AI evaluation, or a closely related field. Strong curiosity and an adversarial mindset, with the ability to systematically push AI systems toward their limits. Experience using structured frameworks, benchmarks, taxonomies, or testing methodologies rather than relying solely on ad hoc experimentation. Strong analytical and documentation skills, with the ability to explain vulnerabilities and risks clearly and precisely. Ability to work independently, adapt to changing project requirements, and move effectively between different AI safety challenges. Knowledge of adversarial machine learning, including jailbreak datasets, prompt injection, RLHF/DPO attacks, or model extraction, is a plus. Cybersecurity experience such as penetration testing, exploit development, or reverse engineering is an advantage. Experience with harassment or misinformation testing, abuse analysis, conversational AI testing, psychology, acting, or creative writing for adversarial scenarios is also valued. Benefits: Fully remote work with the flexibility to complete assignments on your own schedule. Independent contractor engagement with project-based flexibility. Weekly payments through Stripe or Wise based on services rendered. Opportunity to gain hands-on experience in human-data-driven AI safety and red teaming. Direct contribution to making advanced AI systems more robust, safe, and trustworthy. Exposure to cutting-edge AI safety projects and evolving evaluation methodologies. Projects may be extended, shortened, or concluded early depending on project needs and performance. No requirement to access confidential or proprietary information from previous employers, clients, or i