Head of AI Safety

Written by the T&S Network from Moonshot's posting.

Leads applied AI safety work that tests AI systems for pathways to violence, extremism, child sexual exploitation, abuse and grooming, mental health and crisis, and risks to children and teens. The role sits in Moonshot's AI Safety portfolio, serving as the primary applied AI safety counterpart for frontier AI companies, governments, and regulators.

• Child safety and grooming
• Terrorism and violent extremism
• Self harm and suicide
• AI misuse and jailbreaks

What you'd do

• Lead and quality-assure applied AI safety work across harm categories using red teaming and adversarial evaluation.
• Set the methodological approach and turn violence-prevention and safeguarding expertise into testable evaluation frameworks.
• Review test scenarios, model responses, scoring criteria, and evaluation results in detail.
• Spot recurring weak points and edge cases in model behavior, then propose fixes that better protect users.
• Advise frontier AI companies on improving model, product, policy, and intervention safety.
• Manage client and partner relationships, including with governments, regulators, and foundations.
• Lead, coach, and manage the AI safety team, with attention to wellbeing given the sensitive content.

What it asks for

• Several years working in trust and safety, online harms, or an adjacent area like violence prevention, safeguarding, or public health.
• Track record running projects, teams, budgets, partners, and clients.
• Background in winning funding, grants, or procurement work for a practice.
• Excellent written communication for government, foundation, or enterprise audiences.
• Ability to build AI technical fluency quickly and engage with AI company technical teams.
• Prior hands-on work in model safety, red teaming, or adversarial testing of LLMs or AI systems (preferred).

Worth knowing

• Exposure to highly sensitive or graphic content including CSEA, extremist material, and crisis content.
• Wellness support offered, including 24/7 employee counselling access.
• Willingness to travel and work outside regular hours to serve clients or respond to incidents.
• Standard background check and any client-required security clearance.

Terms in this role

CSEA.

Child sexual exploitation and abuse. The wider harm area around CSAM, including grooming, sextortion and trafficking.

Red teaming.

Testing a system, often an AI model, by deliberately trying to make it fail or misbehave, so problems are found and fixed first.

Our best summary of the posting, and it can be wrong. The employer's posting is what counts: read it in full on Moonshotnorth_east before you apply. If we got something wrong, report this listing.