EmergingAI control, monitoring and containment

AI Control Researcher

Also called AI Control Scientist

An AI Control Researcher designs the safeguards that limit what a model can do if it turns out to be unreliable or hostile. They keep risk bounded when you cannot fully trust the system you are deploying.

What they do

They define what a model is allowed to do and what could go wrong if it tried to cause harm. They design protocols that pair a trusted model, a human reviewer or a simple rule with an untrusted model, so bad actions are caught or blocked. They build sandboxes and monitors, then attack their own setup with adversarial evaluation harnesses to find gaps. They write down threat assumptions plainly, run experiments, read the failures and revise the protocol. They also explain the residual risk to engineers and leadership in terms people can act on.

What a strong hire delivers

Within the first months, a strong hire delivers a written threat model for one real deployment, with assumptions stated openly. They ship a working control protocol around it, plus an adversarial evaluation that other engineers can rerun. They give leadership a clear measure of the unacceptable-action rate under those assumptions, and a short list of what the protocol does not cover.

How you can tell they are good

Ask them to take a deployment and state the threat assumptions before proposing any defence. Good candidates name what the untrusted model might try, what the monitor can miss, and how they would test that. Ask for a past attack on their own design. Be wary of anyone who claims zero risk or cannot say what their evaluation fails to show.

Why it is hard to fill

The title is rare and often hidden inside safety, security or evaluation roles. The work needs three things that seldom sit together: security instincts, comfort with model behaviour, and the discipline to state assumptions formally. Most candidates are strong in one or two. Few have shipped a protocol against a real adversarial test, so judging depth takes careful conversation, not a keyword search.

Where to find them

Look in AI safety evaluation teams, security engineering groups that have started working on model risk, and formal methods researchers who have moved towards AI. Frontier labs, safety research organisations and a few security-minded start-ups employ them. Titles are often Research Engineer, Security Engineer, Red Team Researcher or Safety Researcher, so read the work, not the label.

Indicative pay

Not enough published evidence yet to give a fair range for this role, which is common for work this new. We benchmark it for you as part of a search.

Interview questions for an AI Control Researcher

  1. Walk us through a control protocol you designed or studied. What was the threat model and what did it assume about the untrusted model?

    A strong answer states the assumptions clearly, such as whether the model could collude or hide its intent. It explains why the protocol fit them. Worry if the threat model is vague or the answer is only about general safety principles.

  2. How would you measure the unacceptable-action rate for a coding agent, and what would you refuse to claim from the result?

    Look for a defined unacceptable action, adversarial attempts to cause it, and honest limits on the claim. Strong candidates mention sample limits and attacker strength. Worry if they present one number as proof of safety.

4 more questions, with what a strong answer covers, are in the Role Kit.

Job description you can post

We are building AI systems that act with real autonomy, and we want to limit the harm if a model behaves badly. We are hiring an AI Control Researcher to design protocols, test them against adversaries and tell us honestly where the remaining risk sits.

What you will do

  • Define threat models and state assumptions for our deployed AI systems
  • Design control protocols that combine trusted models, monitors and human review
  • Build sandboxing and monitoring around model actions
  • Create adversarial evaluation harnesses and run them regularly
  • Measure the unacceptable-action rate under declared assumptions
  • Write clear reports on residual risk for engineers and leadership

What you will bring

  • Demonstrated work in security engineering, formal methods or AI safety evaluation
  • Experience designing or attacking a defence for a real system
  • Ability to write and test threat assumptions precisely
  • Practical skill evaluating model behaviour with code
  • Clear written communication about uncertainty and limits

Nice to have

  • Experience with formal verification of software or protocols
  • Prior red teaming of AI agents
  • Published or shared research on AI control

Free to use and adapt. If it saves you time, a link back to this page is appreciated.

Skills that matter

  • AI controlAdvancedLimit harm from potentially untrusted AI behaviour
  • Security engineeringAdvancedProtect systems, identities and sensitive assets
  • Model evaluationWorkingMeasure capability and failure modes credibly
  • Formal verificationWorkingProve stated properties under explicit assumptions

Other roles in aI control, monitoring and containment

EmergingAgent Containment EngineerAn Agent Containment Engineer builds the boundaries that keep AI agents from doing harm, even if the model misbehaves or is attacked. They matter…EmergingAI Monitor Evaluation EngineerAn AI Monitor Evaluation Engineer tests whether the systems that watch model behaviour actually catch dangerous or unauthorised actions. It matters…EmergingAI Shutdown and Recovery EngineerAn AI Shutdown and Recovery Engineer builds and tests the ways to stop an AI system mid-action and bring it back to a known good state. It matters…EmergingAI Control EngineerAn AI Control Engineer builds the monitoring, sandboxing and containment that let a company use powerful models safely, even if a model is…

Related roles in other families

EmergingAI Formal Verification EngineerFormal methods and reliable reasoningAn AI Formal Verification Engineer writes precise specifications for AI systems and their outputs, then proves they hold or finds counterexamples…Hiring nowAutomated Theorem Proving ResearcherFormal methods and reliable reasoningAn Automated Theorem Proving Researcher builds and improves methods that find or check mathematical and logical proofs by machine. The company gets…Hiring nowProgram Synthesis ResearcherFormal methods and reliable reasoningA Program Synthesis Researcher builds systems that generate programmes which meet a testable or formal specification. They give the company code…EmergingAI-Assisted Verification SpecialistFormal methods and reliable reasoningAn AI-Assisted Verification Specialist uses AI tools to speed up proof and specification work, then checks every result independently. They help a…

How we wrote this

Role definitions are our own editorial work, checked against occupational classifications, standards, research and employer pages. Titles vary by company, and this is a description of the work, not a job advert.

← AI control, monitoring and containmentAll new job families

Start a search

Hiring an AI Control Researcher?

Tell us the work. We find the people already doing it, whatever their job title says, and you meet them already interviewed for your role.