By Dante D’Orazio
Leaked screenshots of Uber’s customer support platform obtained by BuzzFeed Newsshow that 6,160 support tickets over a 33-month period appear to contain the phrase “sexual assault.” There were 5,827 for “rape.” The data only searches through key data about complaints, such as subject line and driver/passenger name, and not the complaint itself.
Uber defended itself from the leaked data by saying that only five reports of rape and no more than 170 claims of sexual assault from the leaked batch of data were valid allegations. That covers the period from December 2012 through August 2015. The company also published an open letter rebutting the findings.
When questioned about the leak, Uber declined to provide The Verge with a complete number of sexual assault and rape allegations it has received, rather than the subset of complaints highlighted by the leaked data. Instead, a representative noted that the company routinely monitors and manually checks reports submitted by customers to stay on top of such incidents.
The data, notably, only includes reports received through the company’s customer support platform — they do not appear to include reports that went directly to law enforcement, whether or not Uber ultimately found out about the incidents. And complaints about sexual assault or rape that did not include either of those words or variations thereof in the subject line would not be included, either. In all, that means Uber has most certainly received more than just five claims of rape and 170 allegations of sexual assault during the time period in question.
A 24-hour review of the data by Uber led the company to tell BuzzFeed that the figures “significantly overstated” instances of assault across the firm’s services — of all the search hits for “rape,” only five, according to the company, were considered valid. That’s because many different things could provide a keyword to pop up in the search database. For instance, Uber claimed that a passenger named “Don Draper” or expressions using the word could come up as false positives. The company also noted that “rate” is often misspelled “rape” in the complaints. Still, that’s a lot of false positives.
Uber later had to retract claims that any customer or driver name with the word “rape” within it had generated false positives. Zendesk, the company that powers Uber’s report system, said its software would not return customer support tickets in such a fashion. It would, however, group in reports made by any passenger whose name started with the letters R, A, P, and E, regardless of the content of the complaint. Had Zendesk’s search system actually worked as Uber originally stated, many more tickets would have popped up in the BuzzFeed data — up to 28,500 riders and drivers have names meeting that condition, according to company’s estimates.
“We apologize to Zendesk for using an imperfect (and fictitious) example that doesn’t accurately represent their search functionality,” Uber wrote in the updated open letter. “This does not impact our analysis of the overall numbers, which was based on a manual review of these tickets rather than a simple keyword search.”
Even if Uber’s claim of roughly 170 sexual assault complaints and five rape allegations over a 33-month period are taken at face value, it’s a significant number. However, it is difficult to put the figures in perspective, as data for traditional taxi companies is not well-recorded. A number of the largest cities in the US do not break out data into where rape or sexual assault occurred, according to research by The Atlantic from last year.
Other data revealed by the leak shows the company’s internal policies for dealing with such complaints. Service representatives are told to reach out to law enforcement for complaints that cross a certain threashold of concern. If an investigation into non-consensual sexual conduct proves inconclusive, Uber marks the driver with a warning. One more similar warning, and the driver is banned from the service.
More concerningly, Uber’s approach to such sensitive customer complaints takes into account how likely the press or law enforcement (LE) is to be interested in the case. One screenshot reads: “Determine LE/media interest and have Comms/LERT monitor if risk confirmed.” A similar policy appears to be in place for drug and alcohol abuse allegations.
The new findings, however, could raise concerns about the company’s background check procedures, which some have complained are less effective than they should be. The company faced a lawsuit over those checks in 2014, and it is currently fighting a second suit from two victims of sexual assault who say that Uber’s “negligence, fraud and misleading statements” are leading to such incidents.
Update March 6th, 5:42PM ET:This headline and article was modified from its original form to highlight the keyword search data obtained by BuzzFeed. Added link to Uber’s open letter and the company’s decision to decline releasing full allegation data.
Update March 7th, 6:10PM ET: Updated the description of Uber’s defense of its high number of support tickets related to rape and sexual assault due to a retraction the company added to its open letter on Medium.
This article was originally published on The Verge on 3/6/16