AI Red Team Specialist
£133,000–£196,000/year (£11,083–£16,333/month) — Negotiable
Job Description
Test AI systems for abuse cases, unsafe outputs, policy gaps, and operational failure modes. The role turns adversarial testing into reproducible findings that product, safety, and engineering teams can fix.
Requirements
- Experience with AI red teaming, abuse testing, trust and safety, or security testing
- Ability to build test cases, evidence, and severity assessments for model behaviour
- Familiarity with prompt attacks, data leakage risks, jailbreaks, and mitigation validation
- Strong writing for clear technical reports and follow-up tickets
- Ability to build test cases, evidence, and severity assessments for model behaviour
- Familiarity with prompt attacks, data leakage risks, jailbreaks, and mitigation validation
- Strong writing for clear technical reports and follow-up tickets
Responsibilities
- Design and run adversarial tests for models and AI workflows
- Document findings with reproduction steps, impact, and recommended mitigations
- Validate fixes with repeatable test harnesses
- Collaborate with AI safety, security, product, and engineering teams
- Document findings with reproduction steps, impact, and recommended mitigations
- Validate fixes with repeatable test harnesses
- Collaborate with AI safety, security, product, and engineering teams
Benefits
- Work on AI risk issues that affect real product decisions
- Build practical testing methods rather than one-off demonstrations
- Collaborate across safety, security, and engineering functions
- Build practical testing methods rather than one-off demonstrations
- Collaborate across safety, security, and engineering functions