AI Policy Researcher, User Wellbeing, Trust and Safety

About the team

The Trust & Safety Responsible AI Policy group works to make GenAI models and applications safer, fairer, and more dependable. The team does this by shaping policy frameworks, assessing model risk, and creating upstream policy solutions that reduce safety and fairness issues.

This group is part of the T&S GenAI and Emerging Products pillar and partners closely with product policy, product, engineering, data science, operations, red teaming, and other cross-functional teams such as communications, legal, and public policy across global markets. Strong judgment, policy expertise, analytical thinking, creativity, and the ability to explain GenAI risks clearly to different audiences are key to success in this role.

About the role

As an AI Policy Researcher, you will support the responsible development and release of frontier AI models across multiple business areas, with focus on model bias, political risk, and model behaviour. The role combines research, policy design, and launch support to improve how AI systems are evaluated, governed, and monitored.

Responsibilities

• Create and maintain multimodal GenAI policies covering safety-sensitive harm areas such as adult nudity and sexual activity, graphic or shocking material, consent-related sensitivity, mental health, and related risks.
• Turn risk and harm analysis into precise behavioural rules, evaluation standards, grading instructions, and system safeguards.
• Set workable limits between helpful AI use cases and uses that could enable harm, exploitation, misuse, or other unsafe outcomes.
• Develop policy assets that help with model training, evaluation, and deployment, and work with researchers, engineers, product teams, and other partners to put them into practice at scale.
• Build workflows from policy creation through pre-launch testing to post-launch monitoring, including golden set creation, labeling guidance, calibration, adjudication, and evaluation coverage review.
• Use red-team findings, live deployment data, model mistakes, over-refusals, under-refusals, and difficult edge cases to refine policy and measurement quality.
• Spot new capability areas where advanced AI could introduce fresh safety, fairness, or bias concerns, or make harmful behaviour easier.
• Track post-launch model activity to find gaps in the policy framework and improve coverage of unsafe behaviour.
• Advance research that strengthens the justification and practicality of policy decisions, including collaboration with outreach and partnerships teams to gather external expert input.
• Balance long-term safety research with hands-on launch and deployment work.
• Contribute to safety reports, policy documents, launch reviews, and AI governance reviews on responsible AI development.
• Support regulatory teams as a subject matter expert on AI compliance-related work.

Qualifications

The role calls for at least 3 years of experience in Trust & Safety, AI safety research, AI ethics, technical AI governance, or a closely related field. A background in Computer Science, Human-Computer Interaction, Engineering, Data Science, or quantitative Social Sciences is required.

You should also have direct hands-on experience with policy development, AI evaluations, red-teaming, or AI governance. A strong grasp of LLMs, multimodal models, or generative media systems, including their failure modes and safety risks, is important.

Experience working with external experts and stakeholders such as civil society groups, government bodies, and academia is expected, along with a track record of working in a fast-moving technology company or research environment on AI impact assessments, risk assessments, algorithmic audits, data science, or product development.

Excellent written and verbal communication is needed, plus the ability to advocate for safety with stakeholders across Product Policy, Engineering, Public Policy, Legal, Communications, and Data Science.

Preferred experience

It is an advantage if you can explain technical topics clearly to non-technical audiences, have worked with governments, frontier AI companies, or AI safety organizations, are familiar with Python, have built ML systems, and are comfortable working from research through to deployment in production.

Trust & Safety context

This work may involve viewing content in images, video, and text that reflects normal everyday life, but it can also include bullying, hate speech, child safety issues, depictions of self-harm or harm to others, and harm to animals. Candidates should be aware that daily exposure to harmful material is possible in this role.

The company notes that protecting the platform is demanding work that can be emotionally and psychologically challenging for some people. To support employees, comprehensive, evidence-based wellbeing programmes are provided throughout the employee journey, with a person-centred approach built through collaboration across functions.

About the company

TikTok is a leading short-form mobile video platform whose mission is to inspire creativity and bring joy. Its global headquarters are in Los Angeles and Singapore, with additional offices in New York City, London, Dublin, Paris, Berlin, Dubai, Jakarta, Seoul, and Tokyo.

Why join

The company says its products are designed to help people express themselves authentically, discover new content, and connect with others. Teams work with curiosity, humility, and a drive to make an impact in a rapidly growing technology company. The culture emphasizes resilience, experimentation, and an “Always Day 1” mindset.

Diversity and inclusion

The organization says it is committed to building an inclusive workplace where employees are valued for their skills, experiences, and distinct perspectives. It aims to reflect the many communities its platform serves and to celebrate diverse voices across the company.