Remote AI Safety Red Team Tester

Alignerr; France; Sur place; EUR 34 000 - 55 000; Temps partiel

Résumé du poste

Alignerr is seeking AI Red Team Testers to probe, challenge, and outwit AI systems, uncoverting blind spots before harm occurs. Remote hourly contracting role with 10–40 hours per week offers flexible scheduling and meaningful safety work.

No security background required; just curiosity, a systematic approach, and clear reporting. Join a growing field at a pivotal moment in AI safety research.

Qualifications

• Curiosity-driven mindset with a love for puzzles and unconventional ideas.
• Ability to think like an adversary and identify what others miss.
• Detail-oriented documentation of findings in clear, reproducible ways.
• Strong written communication to articulate issues precisely.

Responsabilités

• Design inventive prompts, scenarios, and strategies to expose weaknesses in AI systems.
• Attempt to elicit unsafe, biased, or incorrect outputs from AI models.
• Document failure modes clearly so research teams can address them.
• Assess issue severity using structured evaluation frameworks.
• Collaborate asynchronously with safety and research teams for model improvements.
• Explore edge cases across topics, tones, and contexts.

Connaissances

Prompt engineering

Adversarial testing

Documentation

Structured evaluation

Description du poste

Alignerr is seeking AI Red Team Testers to probe, challenge, and outwit AI systems, uncoverting blind spots before harm occurs. Remote hourly contracting role with 10–40 hours per week offers flexible scheduling and meaningful safety work.

No security background required; just curiosity, a systematic approach, and clear reporting. Join a growing field at a pivotal moment in AI safety research.