Arabic AI Red-Team Analyst (Remote, Türkiye)

As an Arabic AI Red-Team Analyst working remotely from Türkiye, you'll deliberately probe AI models with adversarial prompts to surface safety, bias, hallucination and policy failures in Arabic, Modern Standard Arabic and regional dialects, feeding directly into Corpshore AI's red-teaming and evaluation service line for AI safety-focused clients.

Key responsibilities

• Design and execute adversarial prompts to surface safety, bias or policy failures in Arabic model outputs
• Document discovered failure modes with enough detail for model-training teams to reproduce and fix them
• Work across Modern Standard Arabic and assigned regional dialects depending on project scope
• Flag high-severity findings immediately per project escalation protocol
• Maintain thorough, auditable documentation of all red-teaming sessions

What we look for

• Native or near-native Arabic (MSA required; regional dialect fluency a plus) with strong English
• Strong critical thinking and a structured, creative approach to adversarial testing
• Comfort engaging with sensitive or edge-case content within clear safety guidelines
• Reliable internet and a self-managed remote work setup in Türkiye
• No prior AI safety experience required, full paid training on red-teaming protocols provided

Nice to have

• Background in content moderation, trust & safety, journalism or research
• Fluency in a specific Arabic regional dialect (Gulf, Levantine, Egyptian, Maghrebi)
• Familiarity with generative AI/chatbot tools as a user

Compensation and benefits

• Competitive hourly contract pay, paid via Jwuma timesheets
• Fully remote from Türkiye with flexible scheduling
• Paid onboarding on red-teaming safety protocols and escalation procedures
• Ongoing project flow across Corpshore AI's AI safety and evaluation programs

Common questions