Samsung's introduction of ChatGPT has been "backstabbed"? Experts propose to regulate AI like atomic energy
Samsung introduced ChatGPT for less than 20 days, and three incidents of semiconductor confidential information leakage occurred. Subsequently, employees were required not to use generative artificial intelligence in the workplace; Due to data privacy issues, Italy is investigating ChatGPT for allegedly violating data collection rules and temporarily restricting OpenAI from processing Italian user data.
Artificial intelligence, blockchain and other cutting-edge technologies are accelerating their integration with various industries and daily life, while also facing the current situation of "half is seawater, half is flames". Even neutral technologies, if used improperly or abused, can lead to many problems such as privacy leaks, financial fraud, and false information.
"The Overall Layout Plan for the Construction of Digital China issued this year will jointly list building a credible and controllable digital security barrier and building a self reliant digital technology innovation system as two key capabilities to strengthen digital China." At the 2023 Bund Conference Opinion Forum, He Jifeng, an academician of the CAS Member, said that this shows the key role of digital security in the construction of digital China.
Currently, Shanghai is making greater efforts to promote the construction of a technology and finance center with global influence towards higher levels. Finance and technology are intertwined and mutually reinforcing, exploring a wider range of value spaces.
Li Anmin, Vice President of China Telecom Research Institute, submitted a proposal to build an international digital asset capital last year as a member of the Shanghai Municipal Committee of the Chinese People's Political Consultative Conference. He believed that the earliest feature of the Internet was information sharing, then knowledge sharing, and it has developed into value sharing. "The US dollar system is a comparison of value, and assets are the core element of value. Digital assets manifest as the fusion of virtual and real assets, and the comparison of virtual and real assets will be the core feature of future finance."
Through financial innovation, elevating data resources into productive digital capital can help unleash their intrinsic value. However, risks and foam may coexist. "For example, the protection, security, quality control risks of source data, as well as compliance and consumer protection." Yang Tao, deputy director of the National Finance and Development Laboratory, believes that it is necessary to explore and build a comprehensive risk governance and prevention and control mechanism for digital assets from the macro and micro levels, squeeze the "foam" of capital production, and avoid crowding together.
There are many real-life examples of the importance of building trustworthy AI, such as telecommunications network fraud. The popularization of intelligent terminals has accelerated the speed of people obtaining new information, but the impact of information flow has also provided sufficient materials for black and gray industries.
In 2022, telecommunications network scams not only upgrade and transform traditional fraud patterns, but also create new scams using social hot topics. With the packaging of hot topics, electronic fraud sugarcoating, rapidly updated brainwashing scripts, and the accumulation of tricks and risky operations, it is difficult to prevent.
"The current hot topic of Web3.0, characterized by decentralization, has brought greater challenges to regulation. Technologies such as blockchain and P2P encryption have long been used by black and gray industry groups." Hu Yongtao, Deputy Director of the Network Research Center of the Third Research Institute of the Ministry of Public Security, used two words to describe the digital capabilities of the fraud team - shocking.
For example, money laundering, from the 1.0 version of using mobile black cards and bank black cards to build a money laundering "water house", to the 2.0 version of multiple QR code integrated "aggregate payment", to the 3.0 version of false cross-border trade to flush white funds, to the 4.0 version of QR code "running points", and then to the 5.0 version of digital currency "running points", has only been around for a few years, bringing a huge impact to traditional investigation concepts and experience.
"AI is different from previous technologies in that it has the ability to learn quickly, and its core is to carry life. This means that the development of AI may not be entirely as expected by humans." At the forum, experts also emphasized the importance of not only regulating the systems built by AI, but also valuing AI itself.
Due to the high complexity and uncertainty of algorithms, as well as the strong autonomy of model operation, it may lead to "black box" problems and inexplicability. Prejudice and discrimination that already exist in the data may be further solidified by algorithms, leading to biased intelligent decision-making.
Alternatively, the collection, use, and sharing of data may lead to infringement and leakage of privacy. Samsung is a case study where sensitive data input by users can be "remembered" by big language models, leading to data leakage.
"Artificial intelligence is advancing rapidly, bringing risks and hidden dangers, especially with ChatGPT leading generative AI sweeping the world. AI security risks have shifted from future challenges to urgent threats that require joint efforts from government, industry, academia, and research." Li Yuhang, Chairman of the Cloud Security Alliance for Greater China, stated that experts from the organization have proposed to the United Nations Science and Technology Commission that, given the stronger capabilities of highly intelligent AI in the future and the possibility of self-awareness, AI should be regulated in reference to atomic energy.
In July this year, when Li Yuhang participated in the Global Summit on Artificial Intelligence for Humanity, the Secretary General of the United Nations proposed to consider establishing an AI governance body similar to the International Atomic Energy Agency next year, and the high-level advisory committee is preparing. "AI research and development cannot be abandoned due to choking, nor can it sit idly by and ignore safety. The industry needs to control risks on a global scale, and excellent Chinese solutions should become a model for the world," said Li Yuhang.