Provide the source of power for enterprise development

Provide the source of power for enterprise development

Engine vision | face changing era, face can be thrown

All it takes is one photo
Recently, Momo\'s new AI face changing software \"ZAO\" has become popular in wechat moments and app stores. Different from Momo, \"ZAO\" mainly adopts the entertainment route of AI face changing, allowing ordinary users to upload their own photos and replace their own faces with those of stars in film and television clips, thus experiencing the addiction of acting. In the meantime, the user also can make GIF animation expression through this software, and after the user has selected a photo, after making video or GIF do not have to choose again, can change face with one key. According to the official feature description, using the software can: \"use your face to act in classic movies, use your face to compete with idol, and use your face to make popular emoticons on the Internet.\"
In just one day, AI facechanging App \"ZAO\" went from exploding refresh to becoming the center of the storm
It can be said that \"ZAO\" with the advantage of black box technology and fresh, simple and fun mode, easily detonated the stock, its popular effect is obvious. It was also an overnight hit that brought privacy security and infringement issues arising from the software\'s user agreement to the forefront.
The initial user agreement for ZAO requires the right to receive completely free, irrevocable, permanent, sublicensible, and sublicensible photos of a user\'s face. When users use materials, the user agreement stipulates that \"ZAO\" portraiture rights are granted to \"ZAO\", but it is not clear where \"ZAO\" will use the user\'s facial features and related video, which is equivalent to playing a software to \"sell\" their face for free. If not properly kept or used for illegal purposes, it will give users a great risk of information security.
Recently it was reported that WeChat stopped sharing links to the app due to security concerns. However, the description of sublicensing has been removed from the new user agreement.
When advanced technology meets the dark side of human nature
AI face changing is not new. It has been around since before ZAO, and is best known for Deepfakes. The app has been popular for a while in foreign countries, but some netizens have accused it of being used in the porn industry. By selling video, which is made by AI, celebrities\' faces can be randomly \"grafted\", making many public figures victims. In addition, many sellers offer \"customized face changing\" services on xianyu. Buyers provide celebrity or private photos, sellers can be obscene video protagonist \"graft\", customized face change video price for 1 minute 20 yuan ~50 yuan.
Data abuse times, data encryption, visible unavailable
It is also difficult to define the right to face portrait. Here, it must be pointed out that \"ZAO\" is far from the only App with hidden risks of privacy and portrait right, as the right to portrait and privacy on the Internet has been in a fuzzy zone for a long time. But because of the refresh of \"ZAO\", these problems emerge in the public like an iceberg overnight.
In fact, in the Internet world, users are already running naked, fearing not only that their data will be misused, but also that their information will be altered for illegal use. In fact, these privacy issues are solvable, the common way is to apply cryptography encryption processing, so that data can be used invisible, multi-sharing. This has to mention blockchain, blockchain technology can not tamper with the characteristics of copyright protection, right confirmation, preservation, prevention of infringement. The corresponding cryptography can also solve the problems existing in the field of privacy. Ensuring data security, traceability and data right confirmation are the most important features of blockchain technology.
Privacy computing is based on cryptography algorithm, through secure multi-party computing, proxy reencryption, homomorphic encryption, zero-knowledge proof and other technologies to achieve the purpose of protecting data privacy and security.
Take secure multi-party computing as an example. Under the premise that local data will not be collected and privacy will not be disclosed, without relying on a third party, all parties can still achieve a common desired result by performing the given logical operation. In this process, local and personal sensitive data will not be leaked.
Multi-party computing the MPC allows a group of people to perform joint calculations based on their inputs without requiring each person to display their input values. For example, Alice and Bob want to know who owns more bitcoins, so they can do so without having to disclose how much they own. Unfortunately, the limitation of multi-party computing is that it is extremely inefficient in practice.
The homomorphic encryption Fully homomorphic encryption allows people on encrypted data calculation. This has been an open question in cryptography for decades, until 2009, when Stanford doctoral student Craig Gentry constructed the first fully homomorphic encryption scheme using \"ideal lattices.\" If Bob wants to perform arbitrary calculations on Alice\'s data, such as training the machine learning model, without requiring Alice to display plaintext data, the ideal lattice encryption scheme can come in handy. All - homomorphic encryption and multi - party computing, is still basically in the theoretical stage, the use of efficiency in practice is too low.
The trend of protecting privacy rights is irreversible
The process of using blockchain to register copyright is the same as that of registering in the copyright office, but the efficiency has been greatly improved, and the \"untampered\" feature of blockchain makes the existence of copyright more reliable. On the one hand, publishing copyright notices to declare originality can play a deterrent effect. On the other hand, since information is linked up, infringement is easier to identify and claims are more reliable.
Good technology is like blockchain, which can be applied and innovated under legal supervision. With the development of society, users will have the need to protect their data privacy. The trend of protecting privacy rights is irreversible. In the future, all kinds of data will be linked, and the data flow will be completed by using blockchain technology. This day will not be far away.