The opportunity
Anthropic's Safeguards organization builds the policies, evaluations, and enforcement systems that define and hold the limits on how Claude can be used. In this role, you'll own our conventional weapons work.
What you'll do
Own and maintain Anthropic's conventional weapons policy, defining the: boundary between the activities our models should and should not support
Build the threat models and evaluations that measure how our models could: contribute to weapons development, including the software and autonomy components that define modern weapons systems, and keep them current as the technology and its misuse evolve
Partner with engineering to turn the policy into model guardrails, detection systems, and enforcement tooling
Serve as the subject-matter expert for escalations involving conventional: weapons content, including rapid response to emerging risks
Build shared understanding of the policy across product, engineering, legal,: and leadership, communicating its reasoning to technical and non-technical audiences
Engage external experts as well as government and industry partners, and: translate that engagement into stronger policy and enforcement
What they're looking for
- A working understanding of machine learning and large language model fundamentals
- Hands-on engineering experience in a weapons-relevant technical domain, such: as systems engineering, robotics and autonomy, guidance/navigation/control, sensors and signal processing, aerospace or mechanical engineering, materials and energetics, or embedded software
- Hands-on experience applying arms export controls (ITAR/EAR) or international arms-control regimes
- Experience building or evaluating classifiers, including LLM-based ones, and: reasoning about precision and recall for rare, high-consequence categories
- Trust & safety or product policy experience at a technology platform