Security Red Teaming Intelligence Expert AI Trainer

The work

Evaluate how people might misuse large language models to plan violence or other government security threats. Review model responses to attempts to bypass safeguards and judge whether the content could create a credible, material real-world threat.

Review the duties and qualifications carefully before applying.

• Simulate malicious uses of AI and assess the model’s responses.
• Judge the security relevance of procedural guidance, executable tactics, and specialized knowledge.
• Use operational, analytic, or planning judgment to assess unclear and sensitive scenarios.
• Review sensitive or graphic material and identify implications for threat detection and response.

What it pays and takes

• Pay: Up to $200 per hour.
• Hours: 20+ hours per week.
• Work arrangement: Remote, asynchronous work with flexible scheduling; part-time contract role.
• Location: Open to candidates in the United States.
• Language: Fluent in English.
• Required experience: Seasoned firsthand experience in intelligence, defense, federal law enforcement, special operations, counterterrorism, cyber operations, or related government security work.
• Relevant background: Experience as an analyst, operator, or planner in areas such as counterterrorism, counterintelligence, asymmetric warfare, unconventional operations, threat finance, targeting, wargaming, or red teaming.
• Requirements: Strong judgment in ambiguous threat scenarios and when reviewing highly actionable guidance; ability to assess sensitive or graphic material for security relevance.
• Helpful background: Advanced study through programs such as SAMS, JAWS, NDU, or civilian programs in public policy, public administration, or security studies. Service academy education paired with real-world operational assignments is also relevant.

How it works

Apply on OpenTrain with your resume, then complete the application on the hiring site.

About AI training work

AI training is human work that helps improve AI systems by reviewing examples and model responses. This role relies on specialized security judgment to assess risks that general reviewers may not be equipped to evaluate.