Definição
A prompt that gets a model to ignore its safety rules, for example through role-play, a hypothetical, or a harmful request split into harmless-looking parts. Red-teamers write jailbreaks so the model can be trained to resist them.
Used in: Red-teaming and safety roles.
Nas vagas abertas agora
26 of the open roles we list mention “jailbreak” in the title or description, on 7 platforms (as of 3 November 2026). Of those that state an hourly rate (9 distinct postings), the median is $70/hr, and the middle half pay $65 to $90.
A maioria em: Alignerr (15), Handshake AI (5), Invisible (2), Claru (1), micro1 (1), SuperAnnotate (1) e mais 1.
AI Red Team AnalystAlignerr·Netherlands·Horário flexível
$15–75/hhá 4 dias
AI Jailbreak & Prompt-Injection Security Expertmicro1·Local não informado·Doutorado
$50–90/hhá 6 dias
AI Red Teamer (Remote)Handshake AI·United States·Período integral
$32–95/hhá 29 dias
LLM Jailbreak Red TeamerClaru·Local não informado
$55–100/hhá 3 meses
AI Safety & Red Teaming SpecialistWeekday·India, United States
$50–90/hhá 4 meses
AI QA Trainer - Freelance ProjectInvisible·Singapore·Doutorado
$6–65/hhá 7 meses
Termos relacionados
Adversarial promptA prompt written to make a model fail: to give a wrong, unsafe or rule-breaking answer.Red-teamingDeliberately trying to make a model produce harmful, unsafe or wrong output, so the developer can find and fix weaknesses before release. ← ID verificationTermo anterior, de A a Z. Labelling →Próximo termo, de A a Z. Glossário completoTodos os termos em uma página, de A a Z.Abrir →