Computer security experts are no longerThe first year of warning about the potential dangers of artificial intelligence. In recent years, many neural networks have been invented, which in a few seconds generate photographs of non-existent people, and they may well be used by fraudsters to deceive Internet users. It seems that the experts' concerns are not unfounded - the Associated Press news agency recently recorded one of the first ever cases of using artificially created photos to spy on high-ranking people.
Profile with an artificial photo of a woman wasfound on the social network LinkedIn to establish business contacts. In 2018, the number of registered users in it exceeded 590 million people, and among them there are many high-ranking officials. For example, there you can find Paul Winfrey - former deputy assistant to Donald Trump on domestic policy, so the social network is an attractive place for all sorts of spies.
How to calculate a fake account?
By creating a fake profile and setting asphotos created by artificial intelligence face of a non-existent person, spies can easily get into trust. Due to the uniqueness of the generated images, it is simply impossible to recognize a spy account using Google Image Search. However, Keir Giles, an employee of the Chatham House think tank, was able to identify a fake account - he received a friend request from a suspicious account and compared the information from his page with real data.
LinkedIn's Katie Jones profile
In the profile of some Katie Jones was seta photograph of a red-haired middle-aged woman, and the biography indicates that she worked for a long time at the Center for Strategic and International Studies. The information turned out to be false - a representative of the analytical center Andrew Schwartz announced that a woman with that name had never worked for them. Also in the profile it was mentioned that Katie has a degree in Russian at the University of Michigan, but such a name was not found in the database of the educational institution either.
Features created by artificial intelligence photos
Fake account also gave the seton her picture. To find signs of artificial generation, Associated Press turned to German artist Mario Klingemann, who has been experimenting with artificial intelligence for several years. He is confident that the portrait of a woman with blue-green eyes, red hair and a mysterious smile was created by the generative-adversarial neural network.
Recognize an artificial photo can beindistinct background, blurred outlines of hair and ears, as well as unnatural stripes on the skin. You can verify this on your own by generating a face with the help of a website-generator of realistic faces created by one of the Uber developers. If you take a closer look, you can see that there are particles of hair and coarse bands on the skin in the photos, and the background is completely blurred and full of artifacts.
Conclusive proof of fake accountit was that after an attempt by the Associated Press to contact him, he was immediately removed. It is known that Katie Jones managed to communicate with 52 people, and some of them shared that they often accept friend requests from strangers. Virtually none of them realized that an account could be spyware, and entering into contact with it they put not only personal information at risk, but also expose other people - the likelihood that friends will accept an application from a person with common acquaintances is extremely high.
Representatives of LinkedIn, in turn, announcedthat they are actively working on blocking fake accounts. During the first three months of 2019, thousands of fake profiles were deleted, but they clearly continue to appear. In general, the company advises simply not to communicate with users who are not credible.
You can discuss this topic in the comments, or in our Telegram-chat. It is also convenient to follow the latest materials on our site!