Two German publishing companies sued Google upon learning that, when their names were searched, the resulting AI-generated overview contained incorrect information about them having committed fraud and being unreliable; the court ruled in the companies’ favor. Our co-director Natali Helberger commented on the case for de Volkskrant, stating how this could create a important precedent for holding tech companies accountable for the outputs of their AI models. She highlighted that the judge made a specific distinction between the content of websites served in search results, which are not Google’s responsibility, and summaries created by Google’s own model, which they are assumed to have control over. If the ruling is not appealed, it would imply that Google would have to more heavily scrutinize the output of its models, which takes a very high amount of resources.
Read the article:
