About the role
Scale Labs, Research Scientist — AI Safety and Alignment
As the leading data and evaluation partner for frontier AI companies, Scale plays a central role in understanding and safeguarding AI systems. Building on this expertise, Scale Labs has launched a new team focused on technical AI safety and policy research, to bridge the gap between frontier AI research and policy, helping policymakers make informed, evidence-based decisions about AI capabilities and risks.
Our research tackles some of the hardest open problems in AI safety: assessing the alignment of frontier models, developing control protocols and monitoring methods, preserving the monitorability, understanding failure modes in multi-agent systems and recursive self-improvement. We collaborate closely with partners across industry, the public sector, and academia and regularly publish our findings. We are looking for talented researchers to join us in shaping this vision.
As a Research Scientist on this team, you will conduct research and build evaluations, tools, and datasets that help us understand and reduce risks from frontier AI systems. For example, you might do any or all of the following:
- Design and stress-test control protocols and monitoring systems that detect and limit unsafe behavior by AI models and agents in deployment;
- Build alignment evaluations that probe for deception, sabotage, scheming, reward hacking, and other misaligned behaviors;
- Study monitorability, including whether a model’s reasoning stays faithful and legible enough to supervise, and how training methods and deployment practices can preserve or erode it;
- Investigate risks that emerge when multiple AI agents interact, such as collusion, coordination failures and steerability;
- Develop threat models and evaluations for AI systems that automate AI research and development, to track progress toward recursive self-improvement and inform preparedness planning;
- Work with government agencies and other labs to scope and design evaluations that measure and mitigate risks from advanced AI;
- Publish research and evaluation methodologies, and write technical reports for policymakers.
Ideally you’d have
- Commitment to our mission of promoting safe, secure, and trustworthy AI as capabilities continue to advance.
- Practical experience conducting technical research collaboratively. You should be comfortable building and instrumenting ML pipelines, building evaluation harnesses, running experiments on frontier models and agents, and quickly turning new ideas from the research literature into working prototypes.