Dark Reading is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them.Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

Threat Intelligence

11/22/2019
04:40 PM
Connect Directly
Twitter
LinkedIn
Google+
RSS
E-Mail
100%
0%

Researchers Explore How Mental Health Is Tracked Online

An analysis of popular mental health-related websites revealed a vast number of trackers, many of which are used for targeted advertising.

Researchers who analyzed a collection of mental health-related websites found a vast majority embed "an impressive number" of trackers mostly used for marketing purposes. More than a quarter embed third parties engaged in programmatic advertising and Real Time Bidding (RTB).

Eliot Bendinelli, a technologist with UK non-profit Privacy International, says the organization wanted data protection agencies to take action because it believed there was a fundamental problem with the tracking industry. Its project began with an investigation into sales in the field of ad tech companies, credit rating agencies, ad blockers, and related organizations, he says.

"We were building a case, and basically we think what they're doing is unlawful," Bendinelli continues. While waiting for agencies to act, the research team wanted to find an example of how tracking is taking place on Web pages where people go to read and share sensitive data.

"We wanted a concrete example of how tracking is happening on websites where you think you are safe, and where you are looking up or exchanging data that is sensitive and personal," he adds. They chose sites related to mental health because, as Bendinelli puts it, people may research mental health conditions online because they aren't yet ready to discuss it in person.

The World Health Organization reports 25% of the European population suffers from depression or anxiety each year, and about half of major depressions are untreated. This means every day, millions of people are researching depression online, whether it's to seek help or support someone in their lives who could be suffering from a mental health condition.

At the same time, the Internet's current business model relies on targeted advertisement to make money, tracking individuals and using their personal data to build accurate ad profiles.

To learn how these trackers follow people to mental health websites, Bendinelli and his research team analyzed 136 popular depression-related websites in France, Germany, and the UK. They used a VPN so the search results would be country-specific and began by searching terms related to depression in each of the countries' languages. They collected websites listed on the first page of search results and using a tool called Webxray, analyzed data including the number of trackers on each website, number of cookies dropped, and other analytics data.

What they found "unfortunately, was not surprising," Bendinelli says. "Google is pretty much everywhere," was a key takeaway, and he wasn't shocked to find 97% of websites analyzed had some form of Google tracker. Ninety percent had Doubleclick, which is a business owned by Google that lets online advertisers and publishers display advertisements on their websites. Right behind Google, in terms of numbers, were Facebook and Amazon trackers, he adds.

"These trackers' sole purpose was to do targeted advertising," says Bendinelli.

Real Time Bidding

More than a quarter of webpages analyzed embed third parties who engage in programmatic advertising and Real Time Bidding, a process by which ad impressions are bought and sold. When someone visits a website, a "bid request" is sent to an ad exchange. This request contains various types of data such as demographical data, location data, and browser history. The ad exchange sends this data to advertisers, who bid for the ad impression as it's presented to the site visitor. The one who bids highest will win the impression and have its ad served to users.

The problem, Bendinelli explains, is many companies are learning information the website visitors may not necessarily want to share. "It could be shared with literally hundreds of companies," he says of bid request data. While it isn't sold, it can be used for targeted ads.

Also concerning was the researchers' discovery that a small subset of websites offering depression tests would directly or indirectly share responses and results with third parties. Of the nine testing websites they found, two were storing test results and one was sending responses to another company not mentioned anywhere on the site, Bendinelli explains.

Researchers followed up with the websites they had flagged. One completely removed the test. "They realized they were doing something wrong," he notes.

Bendinelli and Frederike Kaltheuner, tech policy fellow with the Mozilla Foundation, will present more of these research findings at the Black Hat Europe 2019 conference in a briefing entitled "Is Your Mental Health for Sale?"

Related Content:

Check out The Edge, Dark Reading's new section for features, threat data, and in-depth perspectives. Today's top story: "In the Market for a MSSP? Ask These Questions First"

Kelly Sheridan is the Staff Editor at Dark Reading, where she focuses on cybersecurity news and analysis. She is a business technology journalist who previously reported for InformationWeek, where she covered Microsoft, and Insurance & Technology, where she covered financial ... View Full Bio

Comment  | 
Print  | 
More Insights
Comments
Oldest First  |  Newest First  |  Threaded View
7 Tips for Infosec Pros Considering A Lateral Career Move
Kelly Sheridan, Staff Editor, Dark Reading,  1/21/2020
For Mismanaged SOCs, The Price Is Not Right
Kelly Sheridan, Staff Editor, Dark Reading,  1/22/2020
Register for Dark Reading Newsletters
White Papers
Video
Cartoon Contest
Write a Caption, Win a Starbucks Card! Click Here
Latest Comment:   It's a PEN test of our cloud security.
Current Issue
IT 2020: A Look Ahead
Are you ready for the critical changes that will occur in 2020? We've compiled editor insights from the best of our network (Dark Reading, Data Center Knowledge, InformationWeek, ITPro Today and Network Computing) to deliver to you a look at the trends, technologies, and threats that are emerging in the coming year. Download it today!
Flash Poll
How Enterprises are Attacking the Cybersecurity Problem
How Enterprises are Attacking the Cybersecurity Problem
Organizations have invested in a sweeping array of security technologies to address challenges associated with the growing number of cybersecurity attacks. However, the complexity involved in managing these technologies is emerging as a major problem. Read this report to find out what your peers biggest security challenges are and the technologies they are using to address them.
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
CVE-2014-9720
PUBLISHED: 2020-01-24
Tornado before 3.2.2 sends arbitrary responses that contain a fixed CSRF token and may be sent with HTTP compression, which makes it easier for remote attackers to conduct a BREACH attack and determine this token via a series of crafted requests.
CVE-2015-1525
PUBLISHED: 2020-01-24
audio/AudioPolicyManagerBase.cpp in Android before 5.1 allows attackers to cause a denial of service (audio_policy application outage) via a crafted application that provides a NULL device address.
CVE-2015-1530
PUBLISHED: 2020-01-24
media/libmedia/IAudioPolicyService.cpp in Android before 5.1 allows attackers to execute arbitrary code with media_server privileges or cause a denial of service (integer overflow) via a crafted application that provides an invalid array size.
CVE-2015-2688
PUBLISHED: 2020-01-24
buf_pullup in Tor before 0.2.4.26 and 0.2.5.x before 0.2.5.11 does not properly handle unexpected arrival times of buffers with invalid layouts, which allows remote attackers to cause a denial of service (assertion failure and daemon exit) via crafted packets.
CVE-2015-2689
PUBLISHED: 2020-01-24
Tor before 0.2.4.26 and 0.2.5.x before 0.2.5.11 does not properly handle pending-connection resolve states during periods of high DNS load, which allows remote attackers to cause a denial of service (assertion failure and daemon exit) via crafted packets.