In a special edition, Science features news and research on a range of topics ranging from the right to be forgotten to the impact of drones on privacy.
Credit card study shows how little data is needed to identify individuals
Researchers analysed three months' worth of anonymised credit card transactions from over one million people in 10,000 shops in one country. Names, credit card numbers, shop addresses and time of transactions had been stripped out of the data. The remaining metadata included simply shop type and amounts spent – and a code representing each individual.
Researchers discovered just how easy it was to correlate this data with information about a person taken from an outside source to identify individuals – a 'correlation attack'. 90% of the individuals could be identified if their location on four occasions was known; knowing how much was spent on these occasions (using discarded receipts for example) meant researchers could identify nearly everyone – and their transaction history – with three pieces of information per person.
You can read the report Unique in the shopping mall: On the reidentifiability of credit card metadata here.
Camouflaging your search history
Your web search history reveals an extraordinary amount of personal data which search engines can gather – and share. Revelations about NSA snooping increased the public's interest in anonymous search engines. However, what can users who want to continue to use Google and other 'mainstream' search services do to protect their information?
Researchers at New York University have developed a browser extension that produces dummy search requests to 'drown out' real queries. TrackMeNot creates fake search queries by harvesting phrases from popular websites such as The New York Times. Users can customise their fake content searches and can schedule them to take place when they are actually doing 'real' searches.
This software does not hide real queries, but simply disguises them. Meanwhile researchers at Purdue University have developed a tool that does hide a user’s real search – with the downside that search results become irrelevant.
You can read Camouflaging searches in a sea of fake queries here.
Voiceprints and facial recognition
Two articles consider the privacy implications of the latest developments in facial and voice recognition software.
The demand for voiceprint authentication is 'skyrocketing' but the likely growth in voiceprint databases means giving companies and governments another way to identify individuals without their knowledge. As with other 'tracking' technologies, there needs to a balance between convenience and privacy concerns.
Until relatively recently, facial identification has been a struggle for computers. However, deep learning is helping computers improve their recognition capability.
DeepFace – Facebook’s facial recognition software hopes to alert users they have been identified in any of the 400 million new photos updated to Facebook every day. It is reporting an identification rate of 97% in tests - almost on a par with human facial recognition capabilities. Google is also exploring facial recognition software.
Unmasked - the full article on facial recognition - can be found here. When your voice betrays you is available here.
The EU Right to be forgotten
Professor Abraham Newman of Georgetown University US considers global differences in attitudes and approaches to digital privacy – and highlights the need for international approaches to privacy.
The special issue of Science is available here.