Policy Design Manager, Integrity and Authenticity

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

Policy ownership for AI misuse involving inauthentic behavior, influence activity, surveillance, and profiling of individuals and groups at Anthropic. The work sets policy lines for covert monitoring, facial recognition, voice cloning, and geolocation inference, and aims to stop harmful steps early in the surveillance chain.

• Privacy and doxxing
• Coordinated manipulation
• AI misuse and jailbreaks

What you'd do

• Draw the policy line between harmful covert monitoring and legitimate dual-use work.
• Map the surveillance lifecycle from target selection through covert deployment and define policy at each stage.
• Design and oversee evaluations of model capability for social media monitoring, people-finding, and predictive profiling.
• Draft public usage rules and enforcement training, scoring rubrics, and evaluation protocols for reviewers.
• Check human and automated enforcement calls, and take the hardest integrity and surveillance cases.
• Seek input from civil society, privacy experts, threat researchers, and academics on hard calls like research into surveillance tools.

What it asks for

• Understanding of how fake account networks, bot operations, and attribution work in practice.
• Knowledge of misused monitoring tools like spyware, people-search sites, and OSINT methods.
• General knowledge of global data protection, privacy, and platform regulation.
• Strong policy writing skills for turning threat frameworks into enforceable guidelines.
• Experience designing policies and workflows for human enforcement and ML classifier implementation.
• Long track record working alongside engineering, data science, legal, and threat intel staff.
• Five or more years in policy, T&S, threat intel, or platform integrity (preferred).

Worth knowing

• Exposure to explicit content, including sexual, violent, and psychologically disturbing material, and detailed accounts of privacy and surveillance abuses.

Terms in this role

OSINT.

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

Red teaming.

Testing a system, often an AI model, by deliberately trying to make it fail or misbehave, so problems are found and fixed first.

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.