AI Safety Experts — English & Norwegian | $48-$62/hr Remote
Overview
This remote contract role gives you a chance to stress-test conversational AI systems before they reach the public. You'll use your native-level English and Norwegian to probe models for security flaws, bias, and harmful outputs, then turn your findings into structured datasets and reports. You'll be working at the frontier of AI safety at Mercor, where the philosophy is that the safest AI is the one that has already been attacked. The pay is $48–$62 per hour, and you'll have the flexibility to work from anywhere.
What You'll Do4
- 1Attempt to bypass safety guardrails in AI chat systems using techniques such as jailbreaking, prompt injection, and multi-turn manipulation.
- 2Create high-quality training data by labeling model mistakes, categorizing vulnerabilities, and surfacing systemic risk patterns.
- 3Follow established testing frameworks, taxonomies, and playbooks so your adversarial tests remain consistent and reproducible.
- 4Write clear reports and build datasets that document attack cases and give customers actionable insights.
Requirements5
- 1Native-level written fluency in both English and Norwegian is required.
- 2Hands-on experience with AI red teaming, cybersecurity testing, or socio-technical probing of AI systems.
- 3Comfort working within structured taxonomies, benchmarks, and playbooks rather than relying on ad-hoc testing.
- 4A naturally adversarial mindset — you enjoy pushing systems to their breaking points and finding edge cases.
- 5Strong communication skills to explain risks clearly to both technical and non-technical stakeholders.
Who Should Apply
You're the sort of person who instinctively looks for the flaw in a system. You enjoy thinking like an attacker, and you're equally comfortable writing up clear explanations for both technical and non-technical audiences. You're flexible enough to jump between different projects and customers, and you care about making AI safer for everyone. If you have a background in cybersecurity, adversarial machine learning, or creative fields like writing or psychology — and you're fluent in English and Norwegian — this project will be a great fit.
Salary Insight
The role pays $48.00–$62.00 per hour, based on experience and project requirements.
Location
Required Skills
Application Tip
When you apply, include a short portfolio or detailed examples of past red teaming work — especially any attack cases you've created (like jailbreak prompts or adversarial conversation chains). Show how you structured and documented your findings, not just what broke.
Similar open positions
Explore active roles that match your skills and interests.
Mercor
VerifiedAI Safety Experts — English & Finnish | $48-$62/hr Remote
Mercor is assembling a hand-picked red team of bilingual AI safety experts to stress-test conversational AI models in English and Finnish. This remote, hourly role focuses on probing models for vulnerabilities like jailbreaks, prompt injections, and bias exploits, then turning those discoveries into structured data that helps customers harden their AI systems. You'll follow established taxonomies and playbooks, with the option to skip higher-sensitivity projects. It's a unique chance to apply adversarial thinking at the frontier of AI safety while earning a competitive hourly rate.
Mercor
VerifiedAI Safety Expert, Red Team & Adversarial ML
As an AI Safety Expert on the red team, you will probe conversational AI models and agents to uncover jailbreaks, prompt injections, and bias exploits at scale. You will join Mercor, a San Francisco-based talent network backed by Benchmark, General Catalyst, and Peter Thiel, working remotely with a team of elite technical and creative professionals. Your findings will directly shape safer AI deployments for leading research labs. This contract role offers $48–$62/hour and requires fluency in English and Finnish.
Mercor
VerifiedAI Safety Experts — English & Dutch | $48-$62/hr Remote
Mercor is assembling a remote red team of AI safety experts who speak both English and Dutch fluently. Your job is to probe conversational AI models with adversarial inputs, uncover hidden vulnerabilities, and produce the human data needed to make AI safer. This hourly role puts you at the frontier of AI safety, working on text-based projects that tackle sensitive topics like bias and misinformation.
Mercor
VerifiedAI Safety Experts — English & Swedish | $48-$62/hr Remote
This remote, hourly contract is for bilingual (English & Swedish) experts who want to make AI systems safer by attacking them first. You'll red team conversational AI models using adversarial techniques like jailbreaks, prompt injections, and bias exploitation, then turn your findings into structured, reproducible reports. The work is text-based, with optional higher-sensitivity projects supported by clear guidelines and wellness resources. Pay ranges from $48 to $62 per hour.
Mercor
VerifiedAI Safety Expert - Red Teaming
You will red team conversational AI models and agents for Mercor, an AI talent platform backed by Benchmark and General Catalyst. Working remotely and asynchronously, you will identify jailbreaks, prompt injections, and bias exploitation in English and Danish. You will generate high-quality human data by annotating failures and classifying vulnerabilities. Your reports and attack cases will directly improve model performance for leading AI research labs.
Mercor
VerifiedAI Safety Experts — English & Danish | $48-$62/hr Remote
This remote role puts your adversarial skills to work making AI safer. You'll stress-test conversational AI models by probing for vulnerabilities like jailbreaks, prompt injections, and bias exploits. Your findings become the data that helps customers harden their systems. We're looking for native-level fluency in English and Danish, plus a background in red teaming, cybersecurity, or socio-technical risk.