AI Safety Red Teamer, English & Dutch

Probe conversational AI systems with adversarial inputs to identify vulnerabilities, generate high-quality safety data, and help make AI systems more robust and trustworthy. This text-based work includes reviewing model outputs that may address sensitive subjects such as bias, misinformation, and harmful behavior.

• Red team conversational AI models and agents using jailbreaks, prompt injections, misuse cases, bias exploitation, and multi-turn manipulation.
• Annotate failures, classify vulnerabilities, and flag systemic risks.
• Apply testing taxonomies, benchmarks, and playbooks to maintain consistent evaluations.
• Create reproducible reports, datasets, and attack cases that teams can use to strengthen AI systems.
• Expand evaluation coverage by testing more scenarios and uncovering vulnerabilities that automated tests can miss.

• Native fluency in both English and Dutch is required.
• Prior red teaming experience in AI adversarial work, cybersecurity, or socio-technical probing.
• A naturally adversarial, curious approach to testing systems and finding breaking points.
• Experience using structured frameworks or benchmarks rather than relying on unstructured testing.
• Ability to clearly communicate risks to technical and non-technical stakeholders.
• Adaptability to work across projects and customer contexts.

• Adversarial machine learning, including jailbreak datasets, prompt injection, RLHF or DPO attacks, or model extraction.
• Cybersecurity, including penetration testing, exploit development, or reverse engineering.
• Socio-technical risk evaluation, including harassment or misinformation probing, abuse analysis, or conversational AI testing.
• Creative adversarial thinking informed by psychology, acting, or writing.

• Remote, hourly engagement.
• All work is text-based.
• Higher-sensitivity projects are optional and supported by clear guidelines and wellness resources. Topics will be communicated before any content exposure.

$48 to $62 per hour.