Remote Policy & Security AI Evaluation Specialist

AI Trainer Jobs; United States; Remote; USD 62,000 - 83,000; Full time

Job summary

AI Trainer Jobs seeks a remote Security Policy Bypass Security Evaluator to review AI outputs in policy workflows. You will assess citation accuracy, statutory reasoning, and policy adherence, flag risk, and document corrected analyses for training the modeling team.

This independent contractor role requires strong legal/policy credentials, rigorous rubric usage, and the ability to work asynchronously for at least 10 hours weekly.

Qualifications

• Active or recent qualification in policy review or an adjacent practice for Security Policy Bypass Security Evaluator work.
• Hands-on experience drafting, reviewing, or advising on real matters.
• Comfort applying multi-page rubrics consistently across long batches.
• Clear written reasoning that cites authority and names the rule being applied.
• Reliable async availability for at least 10 hours per week.

Responsibilities

• Review AI outputs against current policy review statutes, case law, or policy for Security Policy Bypass Security Evaluator assignments.
• Verify citations, jurisdictions, and dates against authoritative sources.
• Flag risky reasoning, missed exceptions, and policy-adherence gaps with severity tags.
• Capture the corrected analysis or memo paragraph so the modeling team can train on it.
• Adjudicate disputed treatments against published authority or firm policy.

Skills

Legal research

Citation review

Statutory analysis

Policy reasoning

Policy review

Security review

Adversarial testing

Threat modeling

Policy

Bypass

Education

Bar admission or legal-equivalent credential

Job description

AI Trainer Jobs seeks a remote Security Policy Bypass Security Evaluator to review AI outputs in policy workflows. You will assess citation accuracy, statutory reasoning, and policy adherence, flag risk, and document corrected analyses for training the modeling team.

This independent contractor role requires strong legal/policy credentials, rigorous rubric usage, and the ability to work asynchronously for at least 10 hours weekly.