Google CAPTCHA Caught in Privacy Controversy with Blurred House Numbers
As part of new security checks, Google now asks users to read random property numbers snapped by its Street View cameras. The tests known as CAPTCHAs are designed to weed out "bots" by ensuring that users are human. However, it has raised privacy concerns.
Google has been accused of adding the information to its own mapping system, thus exploiting the data submitted by users for commercial gain, reports the Telegraph. According to several campaign groups, the use of such pictures of real house numbers presents "serious" security issues. The groups have accused Google of being "underhand and crude."
The pictures of house numbers were taken from doors and fences on Google's Street View mapping service. When users are asked security questions to access their accounts, Google's Web pages display CAPTCHAs with such blurred house numbers as part of the security checks.
The same image is displayed to numerous Google users worldwide, at the same time. If enough people submit the same answer to the CAPTCHA, Google considers they have accurately read the photo and are therefore human users, not bots. These security checks are estimated to about 200 million a day, and have previously consisted of typing blurred letters into a CAPTCHA box.
Serious Privacy Concerns
According to Nick Pickles, director of privacy and civil liberties at UK's Big Brother Watch, the use of pictures of real house numbers raises serious security concerns, reports the Telegraph. "There is a serious privacy issue with identifying the individual number of people's homes," said Pickles, who has also accused Google of using the pictures to serve its own interests.
Security questions that involve typing a blurred word into a CAPTCHA box are actually part of the search giant's Google Books project. With this project, Google aims to digitalize thousands of physical books that were written before the computer age. When users are asked to type in blurred words, they are actually helping Google to translate scanned images from old, deteriorated books, where the ink has bled or faded, thus making words unrecognizable for a computer program.
No Public Interest, Just Google's
Pickles dismissed any "public interest" in retyping house numbers, and instead claimed Google uses the identification of a house number for its own interests, more specifically to sharpen up the image on its Google Maps or Street View service. "It is clear that Google sees the people who use its services as a commodity to be used up. To use the public as unwitting data loggers is both underhand and crude," he said. "The 'don't be evil' mantra appears to have been replaced with a thirst for knowledge."
According to the Telegraph, a spokesman for Google confirmed that the company has launched a global trial using house numbers as security questions, and noted that such pictures are used in just 10 percent of all questions. All pictures are taken from public roads and highways and the ones with numbers are cropped very closely, therefore they pose no security risks, added the spokesman. He also confirmed that Google will sharpen up the online image once users type the number in correctly.
Google Spokesman Counters Accusations
"We are currently running an experiment in which characters from Street View images are appearing in CAPTCHAs. We often extract data such as street names and traffic signs from Street View imagery to improve Google Maps with useful information like business addresses and locations," said the spokesman, as cited by the Telegraph.
"We take privacy very seriously and the imagery used in this experiment does not contain full details of addresses. We only use images of numbers, street names and traffic signs taken on public property," explained the spokesman. "There is no geographical, individual or contextual information attached to the images whatsoever."
Google has been under fire over security concerns for a while now, and the FCC recently ordered a $25,000 fine after its probe into the Street View cars' data collection case, claiming the tech giant has impeded its investigation.
(reported by Alexandra Burlacu, edited by Dave Clark)