The court said Google must ensure terms generated by auto-complete are not offensive or defamatory. The court case was started by an unnamed German businessman who found that Google.de linked him with "scientology" and "fraud".
Google must now remove defamatory word combinations when told about them, said the court. A person’s privacy would be violated if the associations conjured up by auto-complete were untrue, the federal court said in a statement about the ruling.
However, it added, this did not mean that Google had to sanitise its entire index. "The operator is, as a basic principle, only responsible when it gets notice of the unlawful violation of personal rights," it said.
The ruling on auto-complete overturns two earlier decisions by lower German courts. In the past, Google has defended itself by arguing that it has no control over the combinations of words that auto-complete suggests. Instead, it said, these were automatically generated by the frequency with which other people were looking for such keywords.
A Google spokesman said he was "disappointed and surprised" by the court’s decision, in an interview with Bloomberg. He said it was "incomprehensible" that Google was going to be held liable for the searches carried out by users.