Biological Safety Research Scientist
Anthropic · San Francisco, CA · Safeguards (Trust & Safety)
About this role
Anthropic is hiring a mid-level Research Scientist in the machine learning function based in San Francisco, CA. The posting calls out experience with Python, LLMs, Security, Machine Learning. Compensation is listed at $300,000–$320,000 per year.
- Role
- Research Scientist
- Function
- machine learning
- Level
- mid
- Track
- Individual contributor
- Employment
- Full-time
- Location
- San Francisco, CA
- Department
- Safeguards (Trust & Safety)
More roles at Anthropic
Job description
from Anthropic careersAbout Anthropic
Anthropic’s mission is to create reliable, interpretable, and steerable AI systems. We want AI to be safe and beneficial for our users and for society as a whole. Our team is a quickly growing group of committed researchers, engineers, policy experts, and business leaders working together to build beneficial AI systems.
About the Role
We are looking for biological scientists to help build safety and oversight mechanisms for our AI systems. As a Safeguards Biological Safety Research Scientist, you will apply your technical skills to design and develop our safety systems which detect harmful behaviors and to prevent misuse by sophisticated threat actors. You will be at the forefront of defining what responsible AI safety looks like in the biological domain, working across research, policy, and engineering to translate complex biosecurity concepts into concrete technical safeguards. This is a unique opportunity to shape how frontier AI models handle dual-use biological knowledge—balancing the tremendous potential of AI to accelerate legitimate life sciences research while preventing misuse by sophisticated threat actors.
In this role, you will:
- Design and execute capability evaluations ("evals") to assess the capabilities of new models
- Collaborate closely with internal and external threat modeling experts to develop training data for our safety systems, and with ML engineers to train these safety systems, optimizing for both robustness against adversarial attacks and low false-positive rates for legitimate researchers