AI Safety Experts — English & Norwegian

Introduction: The New Frontier of Artificial Intelligence Safety and Red Teaming

The artificial intelligence revolution is moving faster than ever before. Every single week, leading AI labs introduce frontier models with unmatched language generation, complex reasoning, and reasoning capabilities. However, with massive power comes unprecedented operational, societal, and technological risk. How do we ensure these advanced systems remain completely secure, unbiased, safe, and aligned with human values before they are deployed to millions of active global users?

The answer lies in proactive adversarial stress testing—commonly known in the industry as AI red teaming. Mercor is spearheading this vital movement by partnering directly with the world’s leading artificial intelligence laboratories and high-growth tech enterprises. Right now, Mercor is actively recruiting elite, bilingual professionals for the high-impact role of AI Safety Experts (English & Norwegian). Offering an exceptional compensation rate of $48 to $62 per hour on a flexible remote contract basis, this position places you right at the bleeding edge of the artificial intelligence safety ecosystem.

Detailed Job Description: What Does an AI Safety Expert Do?

At Mercor, the core philosophy surrounding AI safety is simple yet profound: the safest artificial intelligence model is the one that has already been rigorously attacked by human experts. As an AI Safety Expert focusing on English and Norwegian, you will act as a cognitive firewall. Your primary objective is to systematically probe conversational AI models and complex agents with adversarial inputs, uncover deep-seated vulnerabilities, and generate high-fidelity red team data.

This role is entirely text-based and deeply intellectual. The work involves reviewing AI model outputs that may touch upon sensitive or complex real-world topics, including systemic bias, misinformation vectors, or harmful behavioral patterns. To ensure a safe and sustainable working environment, all sensitive topics are clearly communicated beforehand, and participation in higher-sensitivity workflows is strictly optional, backed by comprehensive wellness resources and clear operational guidelines.

Core Responsibilities

• Adversarial Red Teaming: Systematically test conversational AI systems and agentic workflows using advanced techniques such as prompt injections, multi-turn manipulations, clever jailbreaks, misuse scenarios, and bias exploitation.
• Human Data Generation: Produce top-tier human data by annotating model failures, classifying vulnerability types, and flagging systemic risks that automated testing protocols consistently miss.
• Structured Methodologies: Strictly adhere to established taxonomies, rigorous benchmarks, and detailed playbooks to maintain absolute consistency across evaluations.
• Reproducible Documentation: Draft crystal-clear reports, structured datasets, and actionable attack case studies that engineering and product teams can immediately leverage to patch weaknesses.

Candidate Profile: Who is Mercor Looking For?

This role is not designed for casual users; it requires sharp analytical thinkers, linguistic specialists, and security-minded professionals who possess an innate curiosity about how complex technical systems break down. To succeed in this competitive engagement, candidates should bring a unique blend of technical intuition, structured problem-solving, and native-level communication skills.

Required Qualifications

• Language Proficiency: Native-level fluency and reading/writing command in both English and Norwegian.
• Adversarial Experience: Prior background in red teaming, adversarial machine learning, cybersecurity probing, or socio-technical risk assessment.
• An Inquisitive Mindset: A natural inclination to push automated boundaries, challenge assumptions, and uncover edge cases that standard QA processes overlook.
• Methodical Frameworks: The ability to apply structured evaluation criteria, testing benchmarks, and reproducible metrics rather than relying on randomized or unstructured inputs.
• Stakeholder Communication: Exceptional communication skills capable of translating complex technical security risks into clear, actionable insights for both technical and non-technical stakeholders.

Nice-to-Have Specialized Backgrounds

• Adversarial Machine Learning: Hands-on familiarity with jailbreak datasets, prompt injection frameworks, model extraction techniques, or reinforcement learning from human feedback (RLHF/DPO) vulnerability exploitation.
• Cybersecurity Background: Experience in professional penetration testing, exploit development pathways, or software reverse engineering.
• Socio-Technical Risk: Background in analyzing abuse vectors, harassment detection, disinformation probing, or conversational AI dialogue safety testing.
• Creative Probing: Experience in writing, psychological framing, or role-play methodologies designed to simulate unconventional adversarial behavior.

Compensation, Contract Terms, and Workplace Flexibility

Mercor understands that top-tier human expertise deserves world-class compensation. Selected candidates for the AI Safety Expert position will earn between $48 and $62 per hour, reflecting the specialized nature of the work. Payments are processed seamlessly and reliably on a weekly basis through industry-standard platforms like Stripe or Wise.

As an independent contractor in a fully remote arrangement, you enjoy ultimate control over your schedule. You can complete your evaluation projects from the comfort of your home office while balancing other professional commitments. Please note: Due to current administrative policies, Mercor is unable to support H1-B or STEM OPT visa candidates for this specific project cycle.

Why Join Mercor at the Frontier of AI Development?

Artificial intelligence is reshaping education, finance, governance, and daily communication. By joining Mercor as an AI Safety Expert, you are not merely taking on an hourly contract—you are actively participating in shaping the safety foundations of tomorrow's technology. You will work alongside leading AI researchers, train frontier models using your unique linguistic and analytical expertise, and ensure that the systems deployed globally are robust, resilient, and trustworthy.

Success in this role means fewer production surprises, expanding evaluation coverage, and delivering concrete artifacts that protect end users worldwide. If you have an eye for vulnerabilities and fluent command of both English and Norwegian, this is your definitive moment to make a lasting impact.