The most important thing about artificial intelligence fraud is technical content. Automatic speech synthesis software can imitate the voice of people you know, and comfort yourself with "inaudible" when making a phone call. AI spilled a face and changed the face of the hero in the video, which made "what you see is what you get" very embarrassing. Humans will cheat and even attack artificial intelligence.
When the camera recognizes a face, a special pattern is attached to the face, which makes the face recognition technology ineffective and unable to identify or identify errors. When people play poker, their foreheads and faces will be fined. Of course, the machine can not only recognize the machine, but also recognize too many posts, which mom can't recognize.
For another example, after an effective malicious attack on the image of a cat, artificial intelligence may misunderstand the cat after a man-made attack. These deceptive applications are very dangerous in autonomous driving. Attacking the car's vision system and recognizing the wrong traffic signs without the driver will lead to traffic accidents and no longer block the sensor. Where does this malicious attack come from? Some methods are to "differentiate" artificial intelligence by making the pixel value of an image become a small change invisible to the human eye.
As we all know, the deep learning technology of artificial intelligence has become the "eye in the air" of the police. In some cases, the ability to accurately recognize facial photos is better than that of humans. However, there are defects in robustness and universality. This defect will lead to different problems in different environments.
The customer credit score of Internet bank is calculated by deep learning technology, and the artificial intelligence deep learning technology of bank credit information system is very sensitive to the credit value of creditors. When buying a loan, hackers rarely try it. Although the transaction amount is small and there is only one transfer, all methods can turn local tyrants into 1 cent.
Malicious attacks have the upper hand because they meet the above two factors at the same time Wealth appeared overnight. The navy changed the credit score through the remittance records of local tyrants. The bank's credit system will immediately identify hackers as high-value and high-value customers.