轻松盗走 2500 万美元 我们该如何防范人工智能造假危机？

In recent months, artificial intelligence has emerged one after another. We frequently see the following words in news headlines: the company code has been leaked, and the software can't be deleted at all. A large model product has been exposed to reveal private conversations, etc. However, a deeper problem may have quietly emerged before the outbreak of the large model craze. Deep forgery technology has emerged, and anyone in the world has been seamlessly spliced into videos or photos that they have never really participated in through deep learning. It led to the popularity of the word "face changing" in 2000, and the potential threat has not been effectively curbed until today in 2000. Artificial intelligence fraud has become a reality, which is worrying. The animated film ghost in the shell released in 2000 described such a story. The artificial intelligence named Puppet Master knows how to impersonate other identities to confuse and control human thoughts and behaviors. This ability makes it untraceable in the network and virtual space. Although this kind of plot only appears in movies, if you have been deeply loved by Elon Musk or for a long time, The warning influence of Sam altman and others will inevitably assume that once artificial intelligence is left unchecked, it will turn into a skynet threat to human beings. Fortunately, this concern has not been widely recognized by the vast number of artificial intelligence developers. Many academic artificial intelligence giants such as Turing Prize winners have revealed their disdain for this idea in interviews. At present, we do not have artificial intelligence that can drive safely on the road, let alone compete with human wisdom. However, This by no means means that artificial intelligence is harmless and may become one of the most destructive tools created by human beings so far. In this regard, a financial executive who was unfortunately defrauded by artificial intelligence recently realized that his loss was as high as 10,000 US dollars. The case occurred not on the other side of the ocean but in the Hong Kong Special Administrative Region, which is very close to us. Do we have any countermeasures to deal with such artificial intelligence crimes? According to Hong Kong media Radio Television Hong Kong, an anonymous financial controller received a letter from. It is said that the email from his British colleague proposed a secret transaction. The financial controller was not credulous and suspected it was phishing at first sight. However, the sender insisted on arranging a video conference. When the Hong Kong financial controller accessed the conference, he saw and heard the faces and voices of several familiar colleagues. After the doubts were dispelled, they began to prepare the transaction and finally remitted HK$ 100 million in installments according to the instructions of their British colleagues. At present, about RMB 100 million was transferred to a local account until another employee checked with the headquarters. It turns out that the British office has never initiated this transaction and has not received any money. Imagine the extreme regret and guilt of the financial controller after learning that he was cheated. After investigation, it was found that the only real participant in the video conference was the Hong Kong financial controller, and the rest of the staff were the British financial controller and other colleagues simulated by artificial intelligence technology. With only a small amount of publicly available data, a set of basic artificial intelligence programs and cunning deception, these scammers successfully defrauded enough. At the moment, some people may scoff at this story and assert that the financial controller is too careless. After all, the whole live video we usually see on the Tik Tok station is easy to distinguish between true and false, and it is difficult to confuse the false with the true. However, the actual situation may not be so simple. The latest research shows that people can't actually identify it reliably, which means that most people judge what they see and hear by intuition on many occasions. What's more disturbing is that the research also reveals that people are more. It is easy to mistake the video for the real situation rather than the falsehood. In other words, today's technology is smart enough to deceive the audience easily. This word is a combination of deep learning and counterfeiting. Therefore, the core element of technology is machine learning, which makes it possible to quickly produce deeply forged content at a lower cost. If a video creator wants to make someone, he will first use a large number of real video clips of the person to train the neural network to get the appearance of the person from multiple angles and different lighting conditions. Later, the creator combined the trained network with computer graphics technology and superimposed this person's copy on another person. Many people think that a kind of deep learning algorithm called generating confrontation network will become the main driving force of technological development in the future, and the generated facial image can hardly be distinguished from the real face. The first field-specific investigation report devoted a chapter to it, which indicates that they will enable anyone to produce exquisite content. At the end of the year, the user named will be allowed to. Many well-known movie stars and even politicians' faces were used to generate pornographic videos and post them on the Internet. Such fake videos spread all over social networks instantly. Since then, a large number of new videos began to appear, but the corresponding detection technology did not appear until a few years later. However, there are also positive examples that were first known to the audience on the screen because the late actor Paul Walker was resurrected in speed and passion. The New Year's Eve movie Wandering the Earth also reappeared the popular uncle Ng Man Tat with technology, but in the past. It takes a whole studio of experts a year to create these effects. Now this process is faster than before, and even ordinary people who know nothing about technology can create a face-changing video in a few minutes. Concerns about it have led to a surge in countermeasures. As early as in, social media platforms including and banned the use of similar products in their networks, and experts were invited to speak on the stage at major computer vision and graphics conferences. Since criminals have started on a large scale, How can we fight back if we blindly deceive the world? The first strategy of multiple means to defend against risks is to poison artificial intelligence. To create realistic attackers, we need massive data, including numerous photos of the target characters, a large number of video clips showing their facial expressions and clear voice samples. Most technologies use information shared by individuals or enterprises on social media as material, but now there are programs such as and that can modify these files without affecting human perception. However, it is more like a protracted war. With the gradual improvement of artificial intelligence's ability to identify tampered files, the anti-program also needs to constantly develop new methods to maintain its effectiveness. In order to establish a stronger defense line, we need to stop cloning effectively by carrying out such protective treatment on all photos and videos published by ourselves or our enterprises on the Internet. 比特币今日价格行情网_okx交易所app_永续合约_比特币怎么买卖交易_虚拟币交易所平台

注册有任何问题请添加 微信：MVIP619 拉你进入群

文章转载注明 网址：https://netpsp.com/?id=60801