Roles listed here are offered by third-party platforms and clients. Hiring, assessment and onboarding decisions are made by them.
YO AI Labs Live roles
Home/Categories/AI safety & trust AI training jobs

AI safety & trust AI training jobs

17 roles open

There are 17 open AI training and expert data roles in AI safety, trust and red-teaming across the four platforms we refer to. 17 of them publish a rate up front.

The work involves evaluating, correcting and grading AI model output within AI safety, trust and red-teaming — you are paid for professional judgement in your own field, and the roles are remote.

Open roles right now

Mercor: 8Micro1: 6Handshake AI: 2SuperAnnotate: 1
RolePlatformRate
AI Jailbreak & Prompt-Injection Security ExpertMicro1$50-$90/hourApply
AI Safety PractitionerMercor$60 - $70 / hourApply
AI Safety PractitionerMercor$60 - $70 / hourApply
AI Safety Red TeamerMercor$70 - $84 / hourApply
AI Safety Red TeamerMercor$70 - $84 / hourApply
Adversarial Prompt ExpertHandshake AIUp to $65/hrApply
Chemical Safety & Toxicology ExpertMicro1$50-$90/hourApply
Child & Online Safety ExpertMicro1$50-$90/hourApply
Cybersecurity SWE — AI SafetyMercor$60 - $90 / hourApply
Cybersecurity SWE — AI SafetyMercor$60 - $90 / hourApply
LLM Red Team Specialist — Failure Modes & Edge CasesMercor$60 - $90 / hourApply
LLM Red-TeamerMicro1$40-$65/hourApply
Process Safety EngineerHandshake AIUp to $85/hrApply
Public Safety Communications DispatcherMicro1$30-$55/hourApply
Red Team Lead (Offensive Cybersecurity)Micro1$50-$90/hourApply
Red-Teaming Quality Assurance LeadSuperAnnotateUp to $110Apply
Trade Surveillance Expert (Market Abuse & Conduct Monitoring)Mercor$100 / hourApply

What the work involves

Across AI safety, trust and red-teaming, tasks generally fall into a few types: comparing two model answers and justifying which is better, correcting an answer that is subtly wrong, writing hard problems from real practice that models currently fail, and writing reference answers that show the reasoning.

The written justification is the deliverable. Being right matters less than being able to explain precisely why the alternative is wrong.

Who qualifies

Platforms screen on demonstrated practice in AI safety, trust and red-teaming — years of work, tools used, scale handled and the specificity of what you can describe. Most tracks then run an AI-led interview followed by a domain assessment.

How to prepare for the interview →

Other categories

All 30 categories Full live board

Common questions

How many ai safety & trust AI training jobs are open?

17 roles in AI safety, trust and red-teaming are currently open across Mercor, Micro1, SuperAnnotate and Handshake AI, of which 17 publish a rate. The board is rebuilt as listings change.

What do ai safety & trust AI training jobs pay?

Rates vary by platform, seniority and how scarce the specialism is. 17 of the 17 currently open roles in this category publish a rate, and those are shown in the table above.

Do I need AI experience for ai safety & trust AI training work?

No. Platforms are buying your expertise in AI safety, trust and red-teaming. Familiarity with AI tools helps you work faster but is rarely the screening criterion.