• Director Trust Engineering; Aravindh Manickavasagam | Staff Technical Program Manager; Attila Dobi | Sr. • Staff Data Scientist People come to Pinterest to find ideas they feel good about. • To keep that experience safe, we need to know not just what gets reported, but what people actually saw. • That’s what we call prevalence: the percentage of all views, on a given day, that went to content that violates a policy. • Prevalence complements reporting by covering its blind spots, helping us spot under‑reported harms, track trends, and tell whether interventions work. • Why Prevalence Matters Historically, our Trust & Safety teams leveraged multiple indicators to understand the extent of policy violating content on the platform.
Article Summaries:
- Pinterest has introduced an AI‑driven “prevalence” system to monitor policy‑violating content in real time. The platform now samples user impressions daily, automatically labels them at scale, and calculates the share of views that contain violations-providing a statistically robust metric that complements traditional user reports. This approach replaces costly, infrequent human‑review studies, delivering daily prevalence figures with 95 % confidence intervals across policy areas (e.g., adult content, self‑harm) and surfaces (home feed, search). The new radar enables faster detection of under‑reported harms, tracks intervention effectiveness, and supports more timely safety actions.
Sources: