Security Risk Analyst, Risk Engineering

Anthropic builds AI systems designed to be safe, reliable, and interpretable. The company brings together researchers, engineers, policy specialists, and business professionals focused on creating beneficial AI. With offices in San Francisco and New York, Anthropic is reshaping how frontier AI companies approach security and risk.

The Security Risk team at Anthropic operates as an engineering function, handling how the company identifies, prioritizes, and addresses its most critical security risks. You'll help define what modern risk management looks like for an AI-first organization, working directly with CISO-level leadership on decisions that span the entire security landscape. This role sits at the intersection of structured risk analysis, technical security judgment, and strategic business impact.

What you'll do

• Take ambiguous risk questions from leadership and partner teams, drive them to documented decisions, and clearly communicate quantitative and qualitative tradeoffs with honest assessment of uncertainty
• Lead quantitative analysis of top security risk scenarios using FAIR methodology, calibrated estimation, and simulation, working directly with systems owners and translating results for leadership action
• Partner with Security Engineering to evaluate threat scenarios and control effectiveness, ensuring security spending aligns with actual risk and remediation happens in the right sequence
• Frame escalations and risk treatment decisions with clear recommendations, transparent uncertainty statements, and defined re-evaluation triggers
• Shape risk narratives for leadership reviews, help define organizational risk appetite, and establish metrics that matter
• Apply AI and automation to scale risk analysis while owning calibration and quality assurance to maintain trustworthiness
• Convert one-off analyses into repeatable methods, templates, and training materials so partner teams can run increasingly sophisticated risk assessments independently

What you'll bring

• End-to-end ownership of security or technology risk analysis, both qualitative and quantitative, with concrete examples of decisions your work changed
• Hands-on experience with FAIR-style quantification including scenario decomposition, calibrated estimation, and Monte Carlo simulation using Python, R, or spreadsheet tools, presented to decision makers
• Ability to thrive in ambiguity by framing moving questions into analyzable problems, choosing appropriate depth, and driving toward decisions rather than reports
• Technical security depth sufficient to decompose attack paths with engineers while maintaining credibility in those conversations
• Skill in distilling complex risk positions into clear, one-paragraph summaries for executives and defending positions under tough questioning
• Daily proficiency with Claude or other large language models, combined with critical evaluation of model outputs
• Collaborative, low-ego approach that builds credibility through quality work and genuine care about AI safety

Nice to have

• Experience with FAIR-CAM or similar structured approaches to control effectiveness and attack path modeling
• Background building or operating cyber risk quantification programs or converting quantitative analysis into operational tooling and training
• Experience defining risk appetite or tolerance thresholds that organizations actually used for decisions
• Background in security engineering, threat detection, threat intelligence, actuarial science, or decision science
• Familiarity with security risks specific to AI systems, including goal or intent modification and their impact on traditional threat models

What they offer

• Annual compensation of 270,000 to 345,000 USD
• Visa sponsorship available for eligible candidates
• Hybrid work expectation with at least 25% office time at San Francisco or New York locations
• Bachelor's degree or equivalent professional experience required