AI Safety Expert for English and Dutch

Red team conversational AI and generate high-quality human data that exposes vulnerabilities in language models. You will craft adversarial inputs, classify failures, and produce reproducible artifacts that help customers reduce bias, misinformation, and other harmful behaviors. All work is text-based, and you will be informed about sensitive topics before exposure. Participation in higher sensitivity projects is optional and supported with clear guidelines and wellness resources.

• Probe conversational AI models and agents for jailbreaks, prompt injection, misuse cases, bias exploitation, and multi-turn manipulation.
• Review AI outputs that touch on sensitive topics such as bias, misinformation, or harmful behaviors, and annotate problematic responses.
• Generate human data by classifying vulnerabilities, flagging systemic risks, and creating reproducible attack cases.
• Follow established taxonomies, benchmarks, and playbooks to keep testing consistent and auditable.
• Produce reports, datasets, and documented attack examples that customers can act on.

• Prior red teaming experience, for example AI adversarial work, cybersecurity testing, or socio-technical probing.
• Comfort with structured, repeatable testing methods rather than ad hoc approaches.
• Strong written communication skills, able to explain risks clearly to technical and non-technical stakeholders.
• Curious, adversarial mindset with the ability to move across projects and customer contexts.

• Adversarial machine learning, including jailbreak datasets, prompt injection, RLHF or DPO attack techniques, and model extraction.
• Cybersecurity experience such as penetration testing, exploit development, or reverse engineering.
• Socio-technical risk expertise, for example harassment, disinformation probing, abuse analysis, or conversational AI testing.
• Creative probing skills drawing on psychology, acting, or writing to produce unconventional adversarial scenarios.

• You surface vulnerabilities that automated tests miss.
• You deliver reproducible artifacts that strengthen customer AI systems.
• Evaluation coverage expands, resulting in fewer surprises in production.
• Customers gain confidence in the safety of their AI because adversarial scenarios have been exhaustively tested.

• Work at the frontier of human-driven AI red teaming and safety testing.
• Play a direct role in making AI systems more robust, safe, and trustworthy.

• Location: Remote.
• Employment type: hourly.
• All tasks are text-based. Candidates will be notified in advance about sensitive content and can opt out of higher sensitivity projects, which are accompanied by wellness resources and clear handling guidelines.

Pay range: 48 - 62 hourly.

• Native fluency in English and Dutch is required, both spoken and written.
• Ability to work remotely is required. No other work authorization or location specifics were provided in the source.