Facebook's 'People You May Know' recommendation algorithm exposed sex workers' real identities to their clients despite careful separation of personal and professional identities, potentially outing them and compromising their safety.
Facebook's 'People You May Know' (PYMK) recommendation feature caused serious privacy violations for sex workers who maintained separate identities for personal and professional lives. Leila, a sex worker who had only provided Facebook information from her personal identity, discovered that the platform was recommending her regular clients as potential friends, despite having no apparent connections through mutual friends, shared networks, or uploaded contacts. This exposed her real identity to clients and potentially exposed her professional identity to personal contacts. The incident affected multiple sex workers according to secret support groups. Ela Darling, an adult performer, reported similar issues where her professional account was recommended to family members, leading to harassment attempts and forcing her to delete her personal account. Facebook acknowledged using over 100 signals for PYMK recommendations but refused to explain the specific mechanisms or provide effective opt-out options. The company initially claimed users could disable the feature by changing friend request settings to 'No one' but later retracted this explanation, admitting it was incorrect information.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that memorize and leak sensitive personal data or infer private information about individuals without their consent. Unexpected or unauthorized sharing of data and information can compromise user expectation of privacy, assist identity theft, or cause loss of confidential intellectual property.
AI system
Due to a decision or action made by an AI system
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed