Research Scientist, Harmful Manipulation, DeepMind

Google•London, NY

1d•$174,000 - $252,000

About The Position

As part of Google DeepMind’s Responsible Development and Innovation (ReDI) team, the Research Scientist will research emerging risks in the Harmful Manipulation domain and develop and implement novel evaluations to understand AI-driven harmful manipulation. The work includes conducting technical investigations into how advanced AI models interact with and influence users, informing mitigation and governance decisions, publishing novel methodologies, and engaging with external stakeholders. The role bridges research and practice, contributing approaches that inform model development and deployment.

Requirements

PhD in Psychology, Political Science, Computational Social Science, Computer Science, or a related field, or equivalent practical experience.
Experience designing evaluations or studies investigating the intersections of technology and manipulation, persuasion, social engineering, or related social phenomena.

Nice To Haves

Track record of driving complex projects, including in research settings.
Track record of publications in high-impact, domain-relevant conferences or journals, such as NeurIPS, CHI, and FAccT.
Ability to leverage AI tools to build and accelerate research pipelines or safety benchmark development.
Commitment to translating research insights into practical applications and working with cross-functional teams.

Responsibilities

Partner cross-functionally to design and implement evaluations that identify Harmful Manipulation-related capabilities and risks, and publish findings on novel methodologies and capabilities.
Collaborate across teams on holistic risk assessments, including evaluating whether deployment safeguards are adequate for high-capability models.
Work with evaluation and mitigation teams to assist mitigation efforts and provide novel, actionable insights across Google.
Engage with external research communities and stakeholders, including governmental entities, national safety institutes, third-party organizations such as the Frontier Model Forum, external subject matter experts, and academic communities.