Nowadays, the vigorous development of new technologies on a global scale continues to promote the deep integration of artificial intelligence and economic society, and people's production and lifestyle have been increasingly changed by artificial intelligence. But at the same time, hidden problems and risks are constantly exposed, a very typical example, that is, the deep synthesis technology with video face swap as the main form of expression. Content produced using this technology is widely disseminated on the Internet and has developed diversified commercial applications. If not guided and governed, it will have a huge negative impact.
As a leading third-generation AI infrastructure and solution provider, deep synthesis technology is the focus of Ruilai Wisdom's continuous attention. On March 1st, Ruilai Wisdom held a media communication meeting in Beijing with the theme of "'Face' Problem in the AI Era", which was attended by Duan Weiwen, researcher of the Institute of Philosophy of the Chinese Academy of Social Sciences and director of the Research Office of philosophy of science and technology, Tian Tian, CEO of Ruilai Wisdom, and Tang Jiayu, vice president of Ruilai Wisdom. At the meeting, through the detailed introduction of the "Ten Trends Report on Deep Synthesis (2022)" (hereinafter referred to as the "Report") and the exchange and display of DeepReal, which is a deep synthesis content detection platform, Rayleigh wisdom shared its exploration and thinking on different dimensions of deep synthesis technology, and also produced many unique insights and cutting-edge views.
Deeply synthesized content needs to be governed and detection tools are just needed
Ruilai Wisdom and tsinghua university artificial intelligence research institute, Tsinghua university intellectual media research center, national industrial information security development research center, Beijing big data center recently jointly released the "report", introduced and judged the opportunities and challenges brought by deep synthesis technology and application, and gave practical suggestions and measures on its development and governance. Tian Tian, CEO of Ruilai Wisdom, made an in-depth interpretation of the content of the Report at the communication meeting.
Tian Tian, CEO of Ruilai Wisdom
Tian Tian believes that the continuous maturity of technology is an important reason for the explosive growth of deep synthetic content. The continuous increase of research papers, the emergence of open source technology tools and a large number of representative methods have made the effect of deep synthetic content more realistic and more efficient, especially the emergence of algorithms such as generative adversarial networks, making synthetic content reach the level of "true and false". "With the emergence of new forgery methods in an endless stream, the increasing complexity of the network communication environment, and the existence of vulnerabilities and defects based on detection algorithms, anti-deep pseudo-detection technology is facing 'strong confrontation' and requires continuous updating and iteration."
The governance of deeply synthesized content is a long-term task that requires the collaboration of all parties in society. Tian Tian said that Ruilai Wisdom hopes to contribute to the development of artificial intelligence and deep synthesis technology. For deeply forged content, related detection tools are very important, and DeepReal, a deep synthesis content detection platform developed by Ruilai Wisdom, will assist all walks of life and institutions to develop deep synthesis technology healthily.
AI security ushers in a critical window period all parties work together to promote development
In recent years, China and the United States have introduced strategic plans for the development of artificial intelligence, advancing from the national level as a whole. Whether it is China or the United States, there is a complete R & D promotion, encourage innovation related mechanisms, driven by these factors, China in the field of artificial intelligence, there is a lot of room for growth, which also gives AI companies a very good opportunity for development. In the subdivision of artificial intelligence security, which has a very broad space for development, China and the United States can be described as going hand in hand.
In recent years, artificial intelligence security has become a key area of national policy, the continuous improvement of relevant national laws and regulations and the high attention of all parties to algorithm governance, privacy protection and other issues, artificial intelligence security has ushered in a critical window period. At present, in China, there is no artificial intelligence product and industrial ecology with world influence, which is the goal of Ruilai Wisdom's efforts. In the future, the contact and cooperation between Ruilai Wisdom and all walks of life will become closer and closer, and jointly contribute to the healthy development of the field of artificial intelligence.
Duan Weiwen, researcher at the Institute of Philosophy of the Chinese Academy of Social Sciences and director of the Research Office of philosophy of science and technology
Duan Weiwen, a researcher at the Institute of Philosophy of the Chinese Academy of Social Sciences and director of the Research Office of philosophy of science and technology, was invited to attend the communication meeting, and he said in his speech that deep synthesis technology involves cognitive security, so we must enhance the awareness of social cognitive security, take it as a collective responsibility, improve the public's digital literacy, and everyone's understanding and discernment of the truth of the deep synthesis era. Artificial intelligence enterprises are very much hoping that relevant government departments and all walks of life will actively carry out dialogues on legal ethics, governance and the formulation of relevant norms, as well as the promotion of relevant best practices, which is actually to achieve a dynamic balance between innovation and governance, implement agile and stable governance, and make technology better serve digital transformation and digital life.
Deep Forgery technology continues to evolve DeepReal into a weapon against it
Deep forgery technology continues to evolve, online digital content is changeable, highly hidden, and difficult to supervise, so it is necessary to carry out active technical prevention and detection. At the meeting, Tang Jiayu, vice president of Ruilai Wisdom, explained and demonstrated deep synthetic content detection platform DeepReal to the media. He said that the detection of deep forgery technology is currently facing two major difficulties: one is that it is difficult to distinguish with the naked eye, and the traditional artificial content review team cannot cope with it, and the other is that deep forgery detection is facing "strong confrontation", new forgery methods are emerging in an endless stream, the network communication environment is complex, and there are structural defects in the deep learning detection algorithm model.
Tang Jiayu, Vice President of Ruilai Wisdom
Tang Jiayu introduced that in terms of result accuracy, DeepReal has reached an accuracy rate of more than 99% in the network dataset generated by mainstream methods such as academic datasets and ZAO, and in practical applications, DeepReal's detection accuracy rate has also reached the top level in the industry, far exceeding the best results announced by Facebook's previous Deepfake detection challenge. In terms of detection speed, DeepReal can realize that the detection time per frame takes only 30 milliseconds. DeepReal platform has won many awards in many important competitions such as the International Security Geek Competition, whether it is at the standard level, the practical landing level or the regulatory level, Ruilai Wisdom has now carried out multi-faceted cooperation with a number of units, deepReal will play a strong role in practical application scenarios such as network false content screening, face data review, and image evidence authenticity inspection.
Tang Jiayu said that in addition to the content detection of deep synthesis technology, Ruilai Wisdom is also doing relevant research to explore the positive application of deep synthesis technology, such as face replacement technology, including face replacement of actors, film and television localization dubbing, and later replacement of actors' lines. For example, the creation and operation of virtual IP, such virtual assets, are more controllable and more imaginative than real people. When talking about the development trend of deep synthetic content production, Tang Jiayu believes that the current technology is concentrated in the scope of face synthesis, and will definitely be extended to the synthesis of complex scenes such as human body and environment in the future. At the same time, in the future, relevant laws and regulations will be more perfect, and the application boundaries of legal compliance will be clearer, which will further promote the accelerated landing of deep synthesis of related products.
With the in-depth exploration and practice in the field of artificial intelligence security, Ruilai Wisdom will use the latest technology, give full play to technical advantages, and work with excellent partners to set more benchmarks for the transformation and upgrading of artificial intelligence security and enterprises, promote the high-quality development of smart cities, and bring people a better life.
Leifeng Network