Security Risk & Compliance, Agent Security

Security GRC specialist at Anthropic who sets the policy for how AI agents are built, deployed and run safely. The role writes agent governance policies and standards, defines when an agent may touch sensitive or regulated data and which actions need human sign-off, acts as the GRC reviewer inside existing engineering approval workflows, and sets data governance rules such as least-privilege access and access reviews for agents. It also monitors agent behaviour over time, maps agent controls to ISO 27001, ISO 42001 and the EU AI Act, assesses and resolves agent-related risks with engineering owners, and owns the process for risk acceptances and exceptions.

Anthropic wants eight or more years in security GRC, including ownership of policies and control standards at a tech company, experience protecting sensitive data or IP through identity and access design, and clear thinking about non-human identities.