Safeguards Enforcement Analyst, Bio Harms
Enforce Anthropic's Usage Policy against biological and CBRNE harms by investigating model interactions, owning bio-harms monitoring workflows, and partnering with Policy, Threat Intelligence, Data Science, and Engineering teams to improve detection and safeguards. Requires bio-related degree/experience plus Trust & Safety background.
About the job
Key Responsibilities
- Enforce Usage Policies with a specific focus on detecting and mitigating potential bio risks and harmful use of AI systems.
- Own enforcement monitoring workflows for the bio-harms area, improving end-to-end detection, investigation, triage, and escalation processes.
- Monitor and analyze platform activity to identify emerging patterns related to biological (and adjacent chemical, radiological, nuclear, and explosive) threats that may require policy updates or enforcement action.
- Design and architect automated enforcement systems and review workflows that scale effectively while maintaining high accuracy across a technically complex content surface.
- Conduct thorough investigations of potential violations, gathering and documenting evidence to support enforcement decisions.
- Proactively surface trends and propose improvements to detection methods and review workflows.
- Partner with Engineering and Data Science teams to optimize detection models and automated enforcement systems for bio-related policy violations.
- Partner with Policy and Threat Intelligence groups to understand potential exploits and contribute to risk-assessment frameworks, and partner with engineers iterating on safety systems.
- Provide enforcement-grounded feedback on policy gaps, and handle escalations and time-sensitive situations related to potential bio-related Usage Policy violations.
Minimum Qualifications
- Hold a degree in a bio-related field (e.g., microbiology, molecular biology, biochemistry, public health) and/or relevant professional experience in a related field.
- Possess experience in Trust & Safety, content moderation, or policy enforcement at platform scale, including working with generative AI tools to refine and optimize content review and enforcement workflows.
- Possess experience in utilizing AI tools to develop data dashboards for metrics collection in support of continuous improvement efforts.
- Can analyze complex, ambiguous situations and make well-reasoned, defensible decisions under time pressure.
- Are proactive and self-directed: you spot trends, dig in, and ship improvements on your own initiative.
- Communicate clearly in writing and can translate technical bio concepts for diverse audiences, including both technical and non-technical stakeholders.
Preferred Qualifications
- Subject matter expertise in biodefense, biosecurity, WMD/CBRNE non-proliferation, or threat-intelligence.
- An understanding of where real-world biological risk actually lies — adversary intent, acquisition and weaponization pathways, and dual-use concerns, including familiarity with cross-platform threat analysis and open-source intelligence (OSINT) techniques relevant to weapons of mass destruction.
- Proficiency in SQL and/or other data analysis tools to draw insights from large datasets and monitor enforcement workflow health.
- Familiarity with dual-use research of concern (DURC), select agents, and relevant policy frameworks.
Skills
Biosecurity, Trust & Safety, Content Moderation, Policy Enforcement, SQL, Data Analysis, Osint, Threat Intelligence, Generative AI, Risk Assessment
Similar jobs
Fluidstack
Austin, TXBuild and run the global spares sourcing program for critical datacenter equipment (chillers, generators, switchgear) at multi-GW scale. Qualify/dual-source suppliers, set stocking levels tied to failure data and criticality, negotiate VMI/consignment terms, and partner with reliability teams to ensure zero downtime from parts shortages.
Fluidstack
Austin, TXOwn state and local policy agenda for AI data center infrastructure. Build relationships with officials and regulators while drafting testimony and navigating permitting and utility approvals.
Fluidstack
Austin, TXOwn public affairs and community relations for Fluidstack's data center and power infrastructure projects. Build local support through town halls and coalitions, manage media and opposition responses, and partner with internal teams to align external messaging with project realities.
OpenAI
San Francisco, CA
Leads child safety investigations, enforcement decisions, mandatory-reporting workflows, and process improvements involving sensitive content and abuse signals. The role requires Trust & Safety investigation experience, sound judgment, strong documentation, and cross-functional collaboration.
Fluidstack
Austin, TXOwn and maintain Primavera P6 CPM schedules, cost tracking, and productivity reporting for greenfield hyperscale data center construction projects. Drive critical path compression, lead reviews with GCs/executives, and deliver defensible as-built records on $500M+ programs.