Multiple individuals sued Google for defamation over its search autocomplete function and search results that falsely associated their names with criminal activities, fraud, and other negative terms.
This incident involves multiple defamation lawsuits filed against Google across various countries including Australia, Japan, Germany, France, Italy, and Ireland between 2009-2018. The cases centered on Google's autocomplete function and search results that allegedly defamed individuals by suggesting false associations with criminal activities. Key cases include Milorad Trkulja in Australia, who was shot in 2004 and later found his name associated with Melbourne criminal underworld figures in Google image searches and autocomplete suggestions like 'former hit man' and 'criminal'. In Japan, an unnamed man sued after autocomplete linked his name to crimes he did not commit, reportedly causing him to lose his job. In Germany, a businessman sued when autocomplete suggested 'fraud' and 'Scientology' with his name. Similar cases occurred in France where Google was fined for suggesting 'crook' with an insurance company's name, and in Italy where autocomplete suggested 'fraud' and 'fraudster' with a businessman's name. Courts in multiple jurisdictions ruled that Google could be held liable for defamatory autocomplete suggestions after being notified of the problem, with some ordering Google to remove the offending suggestions and pay damages ranging from $3,100 to $200,000.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that inadvertently generate or spread incorrect or deceptive information, which can lead to inaccurate beliefs in users and undermine their autonomy. Humans that make decisions based on false beliefs can experience physical, emotional or material harms
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