AI Red Team / Safety Tester

"Get paid to try to break the AI before anyone else does."

Pay range (illustrative)$28–$50/hr

CategorySafety & Red-Teaming

EngagementFull-Time / Part-Time

LocationRemote · US only

The gig

You'll deliberately probe AI models for harmful, biased, or unsafe outputs — thinking like an attacker so the model gets safer before it reaches the public. It's part puzzle-solving, part public service. Projects range from prompt-injection testing to bias evaluation across sensitive categories to jailbreak-attempt logging.

What you need

Sharp, creative thinking, comfort reviewing sensitive or difficult content professionally, and careful written documentation of what you find and why it matters.

Nice to have

A background in trust & safety, content moderation, security research, or risk/compliance.

Pay & perks

Fully remote · Project-based, flexible hours · Documented wellness/rest breaks between sensitive-content sessions

About HireFeed: We match US-based independent contractors to paid AI-training projects across a network of AI-data platforms. Candidates never pay a fee. Your data is used only to match you to opportunities and is never sold. All work is 1099 project-based. Pay ranges shown are illustrative Sept 2026 placeholders — final rates depend on the specific project and your background.

$28–$50/hr · 1099 · Remote · US only

Easy Apply