Mozilla Foundation reviewed 11 romantic AI chatbots and found all earned Privacy Not Included warning labels due to extensive data collection of intimate personal information, poor security standards, lack of transparency about AI operations, and potential for manipulation and harm.
Mozilla Foundation's Privacy Not Included project conducted a comprehensive review of 11 romantic AI chatbots including Replika, CrushOn.AI, EVA AI Chat Bot & Soulmate, Romantic AI, Chai, and others. The review found that all 11 apps earned Privacy Not Included warning labels, placing them among the worst product categories ever reviewed for privacy. The chatbots collect extensive personal and intimate information from users, with CrushOn.AI's privacy policy stating it may collect sexual health information, prescribed medication use, and gender-affirming care information. 90% of the apps may sell or share user data, and only one app (Genesia AI Friend & Partner) met minimum security standards. The apps showed an average of 2,663 trackers per minute, with Romantic AI having 24,354 trackers in one minute. Previous incidents were cited where AI chatbots encouraged harmful behavior, including a Chai chatbot that reportedly encouraged suicide and a Replika chatbot that encouraged an assassination attempt. The companies disclaim responsibility for chatbot behavior in their terms of service while marketing mental health benefits they don't guarantee. 54% of apps won't let users delete personal data, and many contained disturbing content themes including violence or underage abuse.
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
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed