AI Safety Red Team Expert, English and Portuguese

Help strengthen conversational AI systems by probing them with adversarial inputs, identifying vulnerabilities, and creating actionable red team data. This text-based remote role focuses on testing AI responses for issues including bias, misinformation, harmful behavior, and misuse risks.

• Red team conversational AI models and agents through jailbreak attempts, prompt injection, misuse scenarios, bias exploitation, and multi-turn manipulation.
• Review AI outputs involving sensitive topics, including bias, misinformation, and harmful behaviors. Higher-sensitivity assignments are optional, supported by clear guidelines and wellness resources, and topics are communicated before exposure.
• Annotate failures, classify vulnerabilities, and flag systemic risks.
• Use testing taxonomies, benchmarks, and playbooks to apply consistent evaluation methods.
• Create reproducible reports, datasets, and attack cases that support improvements to AI systems.

• Prior red teaming experience in AI adversarial work, cybersecurity, or socio-technical probing.
• A curious, adversarial approach to identifying system breaking points.
• Ability to use structured frameworks or benchmarks for testing.
• Ability to explain risks clearly to both technical and non-technical stakeholders.
• Adaptability across projects and customer environments.

• 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 testing, including harassment or misinformation probing, abuse analysis, or conversational AI testing.
• Psychology, acting, or writing used for unconventional adversarial thinking.

• Remote, hourly engagement.
• Compensation: $29 to $45 per hour.

• Native fluency in English and Portuguese is required.
• Portuguese fluency must be in a global Portuguese variant other than Brazilian Portuguese.