Safeguards Enforcement Analyst, Integrity & Authenticity

Written by the T&S Network from Anthropic's posting.

Enforcement work against misuse of an AI assistant for coordinated inauthentic behavior, election manipulation, and targeting, tracking, and surveillance of individuals. The role builds and runs enforcement workflows for Anthropic's products and services, covering AI-enabled influence operations, disinformation, election interference, and stalking or profiling. It sits within the Safeguards team and works with Product, Policy, Engineering, and Legal.

• Misinformation
• Coordinated manipulation

What you'd do

• Build enforcement pipelines and review queues that handle volume without sacrificing precision.
• Work with Engineering and Data Science to tune detection models for policy violations.
• Look at flagged content and use the findings to tighten enforcement and policy.
• Act against influence campaigns, fake-account coordination, and efforts to tamper with elections.
• Act against use of the product to watch, locate, or profile people or groups.
• Give the Safeguards policy design team detailed feedback on gaps found in real enforcement cases.
• Follow how threat actors operate and how election, privacy, and surveillance rules are shifting.

What it asks for

• Experience in trust and safety, policy enforcement, threat intelligence, or a related field focused on influence operations, disinformation, CIB, election integrity, or privacy and surveillance harms.
• Has launched and expanded enforcement or content review operations.
• Comfortable querying large datasets in SQL or similar analysis tools.
• Able to spot new risks and threat actors and brief people across product, policy, engineering, and legal.
• Hands-on use of generative AI tools, including crafting prompts for moderation tasks.
• Understanding of implementing product policies at scale, including content moderation.
• Cross-platform investigation experience into influence operations or disinformation (preferred).

Worth knowing

• May be exposed to explicit political, violent, or psychologically disturbing content.
• Responds to escalations during weekends and holidays, especially around major electoral events.

Terms in this role

Coordinated inauthentic behavior.

Networks of accounts working together to mislead people about who is behind them. Meta's term for influence operations, now widely used.

OSINT.

Open-source intelligence: investigating with publicly available information, such as social media, public records, archived pages and imagery.

Our best summary of the posting, and it can be wrong. The employer's posting is what counts: read it in full on Anthropicnorth_east before you apply. If we got something wrong, report this listing.