Safeguards Enforcement Analyst, Ban Evasion & Recidivism

Anthropic’s mission is to create reliable, interpretable, and steerable AI systems. We want AI to be safe and beneficial for our users and for society as a whole. Our team is a quickly growing group of committed researchers, engineers, policy experts, and business leaders working together to build beneficial AI systems.

As a Safeguards Enforcement Analyst on the account abuse team, you'll build and execute enforcement workflows that keep our products safe, with a focus on detecting and mitigating potential harm. Your initial focus will be recidivism: a ban that an actor can evade in five minutes isn't enforcement — it's friction. You'll own detecting when banned actors return, linking accounts across identities, and closing the re-registration paths that matter most. The mandate includes our highest-stakes populations, including preventing evasion of child-safety enforcement bans, where the cost of a missed return is unacceptable.

This position may expand into broader areas of enforcement over time. Safety is core to our mission, and you'll help shape policy enforcement so that our users can safely interact with and build on top of our products in a harmless, helpful, and honest way.

• Investigate evasion clusters end to end — from a single appeal or signal anomaly to the full linked actor network
• Convert individual findings into durable systemic controls and detection proposals
• Operationalize re-registration controls for high-severity ban populations
• Partner with Engineering and Data Science teams on account-linking signals to connect returning actors across identities
• Build the recidivism measurement framework: how often banned actors return, how fast we catch them, and which controls reduce return rates
• Author playbooks for contractor-supported evasion review with QA against your own gold standard
• Keep up to date with emerging AI policy enforcement best practices, and use these to inform our decision-making and workflows

• Experience investigating ban evasion, multi-accounting, or repeat fraud actors at a platform with adversarial users
• Fluency in SQL and comfort building your own analyses across large account and event datasets
• Experience working with fraud or identity-linking signals and a working understanding of their precision/recall tradeoffs
• Rigor about evidence standards — comfort with the asymmetric cost of false positives in severe-harm enforcement
• A track record of turning one-off investigations into repeatable detection logic and policy
• Strong written communication skills, with experience producing clear briefs and recommendations for technical and non-technical stakeholders
• Excellent judgment and the ability to collaborate with team members while navigating rapidly evolving priorities and workstreams

• Experience using payment or network risk signals in an enforcement context
• Experience with child-safety or other high-severity integrity enforcement
• Experience collaborating directly with detection engineering or data science teams on rule deployment
• A deep interest in AI safety and responsible technology development
• Experience writing effective prompts for generative AI systems in a content review or enforcement context