Remote AI Safety Red Team Tester

Alignerr Corp.; Italia; Remote; EUR 25,000 - 74,000; Part time

Job summary

Alignerr is seeking AI Red Team Testers to probe, challenge, and stress-test AI systems, identifying weaknesses before they cause real-world harm. This remote hourly contract offers 10–40 hours per week and the flexibility to work from anywhere.

You will design prompts and scenarios, attempt to elicit problematic outputs, document failure modes, and rate severity. You’ll collaborate asynchronously with safety researchers to help prioritize fixes and improve AI safety at scale.

Qualifications

• You thrive on solving problems and spotting the unexpected.
• Comfortable thinking adversarially, asking what could go wrong.
• Detail‑oriented with clear documentation.
• Strong written communication, concise and precise.
• Self-motivated and able to work asynchronously.
• No cybersecurity or hacking background required.

Responsibilities

• Design prompts, scenarios, and conversational strategies to expose AI weaknesses.
• Attempt to elicit incorrect, unsafe, biased, or problematic outputs from AI models.
• Document discovered failure modes with clear reproduction steps.
• Assess and rate the severity and potential impact of each vulnerability.
• Collaborate asynchronously with AI safety and research teams to prioritize fixes.
• Contribute to making AI systems more robust, fair, and trustworthy.

Skills

Problem solving

Adversarial thinking

Detail oriented

Written communication

Independent worker

No cybersecurity background

Job description

Alignerr is seeking AI Red Team Testers to probe, challenge, and stress-test AI systems, identifying weaknesses before they cause real-world harm. This remote hourly contract offers 10–40 hours per week and the flexibility to work from anywhere.

You will design prompts and scenarios, attempt to elicit problematic outputs, document failure modes, and rate severity. You’ll collaborate asynchronously with safety researchers to help prioritize fixes and improve AI safety at scale.