Product Manager, Safeguards (Account Integrity & Abuse)

About the role

Anthropic is building AI assistants that are helpful, harmless, and honest. The Safeguards team protects users from misuse and ethical risks, and the Product Manager will own the design, development, and deployment of safety systems and related user experiences across cloud platforms.

Key responsibilities

• Determine how to build safety by design upstream and leverage downstream defenses for Anthropic’s frontier models, AI products, and customers on surfaces such as Claude.ai, 1P API, and external cloud providers.
• Write safety evaluations and communicate findings externally.
• Prioritize ruthlessly by defining problems, solution options, and clarifying business and technical trade‑offs to shape MVPs versus ideal states.
• Align and collaborate with policy, enforcement, research, engineering, and other cross‑functional stakeholders.
• Monitor the AI landscape to plan mitigation of deployment risks from increasingly powerful models and determined adversaries.
• Lead development of metrics to surface performance, blind spots, and inform future project planning.

Required profile

• Ability to make technical trade‑off decisions, ideally with experience working across policy experts, AI/ML research engineers, and software engineering teams to design state‑of‑the‑art safety systems.
• Strong understanding of how products are used, associated safeguards concerns, and how to deliver effective solutions.
• Proven ability to build product and engineering strategy across multiple cross‑functional teams in a rapidly changing environment.
• Experience designing and building metrics to evaluate risks, system performance, and user impact, and making clear trade‑offs.
• Exceptional navigation and prioritization skills amid rapidly changing product specifications, with flexibility to bring clarity and execute.
• Evidence of sound judgment and decision‑making in ambiguous situations.
• Track record of planning, building, launching, and measuring new products or systems from zero to one.
• Clear articulation of complex technical concepts to non‑technical audiences, both in writing and verbally.
• Creative thinking about risks and benefits of new technologies beyond existing checklists and playbooks.