Google 'autocomplete' and 'related search' functions reveal identities of sex offence victims in Times probe

Google’s autocomplete and related search functions are revealing the names of sex offence victims in internet searches, according to an investigation by the Times.

The paper searched the names of attackers and alleged attackers in sex assault cases on Google, which then suggested the names of the women they had been convicted or accused of abusing.

They either appeared as a keyword suggestion in the search bar or in the related searches section at the bottom of the results page.

 Loading ...

Searching a victim’s name could also reveal the identity of their abuser, according to the Times.

Under UK law, sex offence victims are given lifetime anonymity, which comes into force even when an allegation is made, precluding the publication of anything that might lead to their identification.

Breaching this is a criminal offence, which can lead to fines of up to £5,000.

Google’s autocomplete algorithm automatically suggests words based on what a user is searching and what others have also searched for using those same keywords.

In some cases victims have already been illegally identified on social media.

The Times said it performed searches from “several unlinked computers” to avoid search history asserting any influence on the results of its investigation.

In one alleged rape case, typing the defendant’s name plus a “common search term” brought up the alleged victim’s name under autocomplete, the Times said.

In a separate alleged rape case, searching a defendant’s name revealed a woman’s name and home town under related searches.

Google has been informed of the breaches.

A Google spokesperson told the Times: “We don’t allow these kinds of autocomplete predictions or related searches that violate laws or our own policies and we have removed the examples we’ve been made aware of in this case.

“We recently expanded our removals policy to cover predictions which disparage victims of violence and atrocities, and we encourage people to send us feedback about any sensitive or bad predictions.”