Facebook's AI content moderation system incorrectly flagged 20 posts from the Auschwitz Memorial Museum as violating community standards for bullying, harassment, and nudity, and deleted one image of orphans.
Facebook's automated content moderation system incorrectly labeled 20 posts from the Auschwitz Memorial Museum as violating community standards. The posts contained historical photographs of Holocaust victims who died at the concentration camp in Poland. The AI system flagged these respectful memorial posts as showing 'bullying and harassment' and 'adult nudity and sexual activity'. One image showing a group of orphans was deleted entirely. The museum described the censorship as 'unacceptable and offensive to the memory of the victims of Auschwitz' and demanded an explanation. Meta, Facebook's parent company, later claimed the photographs were not actually moved down the museum's feed and that the violation notices were sent by mistake. The company admitted that the image of orphans had been deleted by mistake and was subsequently restored. Meta offered a sincere apology for the error. The museum had been posting similar historical content for years without any issues before this incident occurred.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that fail to perform reliably or effectively under varying conditions, exposing them to errors and failures that can have significant consequences, especially in critical applications or areas that require moral reasoning.
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
No population impact data reported.