AI security, safety and governance
Hiring for AI security, safety and governance
Guides to the specialist roles that keep AI systems secure, safe and well governed, from red teamers who attack models before anyone else does to researchers who study how they fail. Each page explains what the role involves, the skills to look for and how it differs from the roles around it.
Our approach
How we hire for AI security
- 01
Define the role
We start with the systems the hire will protect, the threats that concern you most and the team they'll join, then agree what the role needs to cover.
- 02
Search where the specialists are
AI security people sit across research groups, security teams and AI labs. We look at the work candidates have done, including research, tools and assessments, not only their job titles.
- 03
Assess before you meet
We talk to candidates about their work in technical detail before introducing them, and share notes on each person with the shortlist.
01
Offensive security and red teaming
Engineers and researchers who attack models, agents and AI products to find weaknesses first.
AI Red Team Engineer
Hiring the engineers who attack AI systems before anyone else does.
View roleLLM Red Teamer
Hiring specialists who test language models for jailbreaks, leaks and misuse.
View roleAdversarial ML Engineer
Hiring engineers who build and defend against adversarial attacks on models.
View roleAI Penetration Tester
Hiring penetration testers who assess AI applications, agents and model APIs.
View rolePrompt Injection Researcher
Hiring researchers who study how prompts can hijack AI systems, and how to stop it.
View roleAI Offensive Security Researcher
Hiring researchers who develop new offensive techniques against AI systems.
View roleAI Vulnerability Researcher
Hiring researchers who find, reproduce and report flaws in AI models and tooling.
View roleAI Bug Bounty Hunter
Hiring researchers who hunt for vulnerabilities in AI products.
View role
02
Research
Researchers who study attacks on models, harmful behaviour, alignment and what happens inside neural networks.
AI Security Researcher
Hiring the researchers who find and fix weaknesses in AI systems.
View roleML Security Research Scientist
Hiring scientists who publish rigorous research on attacks and defences for ML.
View roleAdversarial Robustness Researcher
Hiring researchers who make models hold up under adversarial and shifted inputs.
View roleAI Safety Researcher
Hiring researchers who study how AI systems can cause harm, and how to prevent it.
View roleAlignment Researcher
Hiring researchers who work on making AI systems pursue the goals intended.
View roleInterpretability Researcher
Hiring researchers who explain what happens inside neural networks.
View role
FAQ
AI security hiring questions
What does an AI red teamer do?
An AI red teamer attacks models, agents and AI products the way a real adversary would, to find weaknesses before release. The work covers jailbreaks, prompt injection, data leakage and misuse of connected tools, and ends with findings that engineers can fix. See our guides to hiring an AI red team engineer and an LLM red teamer.
How is AI security different from cyber security?
Cyber security protects networks, systems and data. AI security covers the same ground plus risks that come from the model itself, including prompt injection, poisoned training data, model theft and agents that misuse the tools they're given. Many AI security specialists come from cyber security and add machine learning knowledge, or the other way round.
What's the difference between AI safety and AI security?
AI security deals with attackers, meaning people trying to break, steal or misuse a system. AI safety deals with harm that comes from the model's own behaviour, including dangerous capabilities and goals that don't match what was intended. The two overlap, and some teams cover both. See our guide to hiring an AI safety researcher.
What should an AI security job description include?
Say which systems the hire will work on, which threats matter most to you and whether the role is research, testing, engineering or governance. Name the frameworks you work to, like the OWASP guidance for LLM applications, MITRE ATLAS or the NIST AI RMF, and say who the role reports to.