Enforcement Analyst, Agentic Alignment
Anthropic · Boston, MA; San Francisco, CA | Seattle, WA | New York City, NY | Washington, DC
About this role
**Enforcement Analyst, Agentic Alignment — Anthropic** ## About Anthropic Anthropic’s mission is to create reliable, interpretable, and steerable AI systems—so AI is safe and beneficial for users and society. ## About the role As an **Enforcement Analyst focused on Agentic Alignment**, you’ll identify and investigate harmful or unexpected model behavior and make enforcement decisions accordingly. You’ll collaborate with cross-functional teams (including Policy, Threat Intel, Engineering, Research, and Transparency) to help maintain Anthropic’s commitment to safe and beneficial AI as Claude takes on more autonomous work. **Important context:** In this role, you may be exposed to explicit content (sexual, violent, or psychologically disturbing). There is also an **on-call** responsibility across the Policy and Enforcement teams. ## Key responsibilities - Review and investigate potential instances of **Agentic Misalignment** using transcripts, tool-use traces, and code. - **Triage** cases by severity and urgency and route them through the appropriate channels. - Help operationalize **review pipelines** for agentic cases, turning detection leads and investigations into scalable workflows. - Make enforcement decisions under the **Usage Policy**, applying it consistently to novel agentic scenarios and documenting your reasoning. - Assess cyber-related agentic activity (e.g., scanning, exploitation attempts, credential misuse, malicious code) to determine real-world harm potential. - Write clear, well-evidenced **case write-ups** for enforcement peers through senior leadership. - Identify **patterns and trends** across cases, and feed findings back to policy, detection, and research teams to improve policies, classifiers, and model behavior. - Maintain rigorous documentation of enforcement decisions, review logic, and precedents. - Collaborate with Threat Intelligence, Research, Engineering, and Policy to keep detection coverage aligned with evolving agentic misuse. - Use AI systems (including Claude) to accelerate investigations (summarizing transcripts, analyzing code, querying data, building lightweight tooling). - Contribute to regular reports on agentic harm trends and enforcement outcomes. ## Minimum qualifications - Can read and understand code (e.g., Python, JavaScript, shell) well enough to follow what happened and why it matters. - Background in security, cyber threat intelligence, or a similar technical field to judge cyber harm severity. - Experience in trust & safety, enforcement, investigations, or related work—translating ambiguous policy questions into defensible decisions and written findings. - Conducted structured investigations from first signal to final finding, knowing when evidence is sufficient. - Supported formal enforcement actions (documentation, evidence packaging, escalation coordination). - Clear writing and ability to present complex findings to stakeholders across Legal, Policy, Engineering (including non-technical audiences). - Sound judgment on escalation needs and resilience working with difficult content. ## Preferred qualifications - Demonstrated use of AI systems to accelerate your work and interest in finding new ways to do so. - Comfortable working with large volumes of data (long transcripts, logs, traces, multi-table datasets) and surfacing meaningful behavioral patterns. - Curious about how AI agents behave and enjoy investigating uncategorized behavior. - Familiarity with AI safety/alignment an
Listing freshness
CronJobs last confirmed this listing 1h ago. If its source stops confirming the opening for seven days, this page is removed from active inventory.