Essential

1. Demonstrated capability breaking or securing AI systems
2. Experience with LLM internals, prompt injection techniques, or model evaluation
3. Ability to work independently and drive projects to completion

‍

Highly valued

1. Successful jailbreaks of frontier models (GPT-5, Claude, Gemini, etc.)
2. Published security research or significant bug bounty findings
3. Professional experience in security consulting, red teaming, or ML security
4. Prior work with AI labs or safety organizations

1. • Remote work
2. Flexible hours, can work around schedule
3. Opportunity to work with frontier AI models
4. Mentorship from experienced AI security researchers

Your work will contribute to making AI systems safer by helping identify and mitigate failure modes before deployment.

‍

What you'll gain:

1. Hands-on experience with frontier AI security
2. High-impact work on cutting-edge AI safety challenges
3. High ownership and autonomy
4. Strong references for future roles in AI safety

1. Submit your application in minutes
2. Play The Specification Game ✨, where you’ll jailbreak a frontier model
3. 30-minute interview with our founders
4. Offer