Verification code login, mobile online payment, face recognition access control. Technology is a double-edged sword. While bringing efficiency improvements to our production and life, security issues cannot be ignored.
With the development of a new generation of information technology, data has become an important factor of production. Today, with the rapid development of the Internet and artificial intelligence technology applications, data is even more regarded as an intangible asset. In recent years, when technology companies promote a new generation of information technology products, issues such as data security and data privacy have been widely criticized. Data security and personal data protection have become key and pressing issues in today’s digital transformation era.
At the 2021 World Artificial Intelligence Conference held from July 8th to 10th, issues such as data security, data governance, and network security have attracted much attention. In the process of deepening the digital transformation of cities with artificial intelligence, the industry should actively embrace supervision and promote the healthy and sustainable development of the industry.
The hidden worries of data security on the road of AI A few days ago, due to serious violations of laws and regulations in the collection and use of personal information in the App, Didi Chuting was removed from the shelves by the State Internet Information Office in accordance with the relevant provisions of the Cybersecurity Law of the People’s Republic of China.
Earlier in the Shanghai Auto Show Tesla owner’s rights protection event, Tesla announced the driving data of the rights protection owner’s vehicle half an hour before the incident. Should auto companies disclose their driving data, how people determine the reliability of data and other issues that have aroused widespread debate in society?
“Big data-driven business means that data security has become unprecedentedly important to the operation of cities. Once data is attacked, it means that the business will stop, causing serious economic losses and social consequences.” Zhou Hong, founder, chairman and CEO of 360, said It was stated at the opening ceremony of the 2021 World Artificial Intelligence Conference that in recent years, 360 companies have received a large number of reports from hospitals in China. After being attacked by ransomware, the hospitals cannot perform operations and cannot register patients for medical treatment, which has a serious impact.
Without the collection and foundation of big data, artificial intelligence is without roots. Data is an important foundation that drives the rapid development of this round of artificial intelligence. The safety of data determines the safety of artificial intelligence. The accelerated development of artificial intelligence has brought new opportunities and challenges to data security governance.
In recent years, cities have become an important target of ransomware supply. The targets of cyber-attacks are not limited to computers, mobile phones, equipment, and systems, but extend to data. The ransom of data from governments and urban public utilities will cause urban operations and services. Stuck into a lockout.
Data security also directly affects national security. In May of this year, the largest fuel pipeline in the United States was forced to shut down due to a cyber-attack, which affected 45% of the supply of gasoline and diesel fuel on the east coast of the United States.
The acquisition, processing, and application of artificial intelligence data have strong privacy and domain scalability and therefore are related to citizens’ personal information security and even national information security. Artificial intelligence technology still has to hang a sword of Damocles above the head. How to ensure that public data is not used for commercial purposes? Not only the public but also a constant concern of domestic public security departments.
Which links are prone to data security vulnerabilities?
Illegal cross-border circulation of data may endanger national sovereignty and national security; excessive collection and illegal use of digital information may infringe the rights and privacy of citizens; algorithm preferences may aggravate social prejudice or discrimination, threatening fairness and justice; driverless cars are urgent Intelligent decision-making such as risk aversion may threaten the lives of certain groups of people.
Chen Zhimin, deputy director of the Social and Legal Affairs Committee of the National Committee of the Chinese People’s Political Consultative Conference, pointed out at the high-end security dialogue of the Artificial Intelligence Conference that the development of artificial intelligence and the risks of security development coexist at the same time. It is not only a technical issue, but also a security issue, but also involves ethics, law and international Rules related issues.
He Xiaolong, deputy director of the National Industrial Information Security Development Research Center, said at the Artificial Intelligence Conference that the data security issues brought about by artificial intelligence are mainly reflected in the acceleration of traditional data security issues on the one hand. The large-scale use of artificial intelligence has further exacerbated the security problem of excessive data collection. On the other hand, it brings a new type of data security problem. Artificial intelligence algorithms have a strong dependence on data, which may bring similar new data security challenges such as data poisoning.
Algorithms, models, and data constitute the three cornerstones of machine learning. Zhu Jifeng, the chief AI security expert of Tencent’s Suzaku Lab, stated on the theme forum of the Artificial Intelligence Conference that from the perspective of the upstream and downstream industrial chain in the algorithm construction process, from the perspective of data input, data analysis, model training, model decision-making, and model development. There are corresponding risks in the online deployment of applications and model transformation.
For example, in the data input link, if the attacker controls the source of the data, or the attacker can launch a security attack on the data-collecting device itself, the algorithm model will eventually be affected through this chain. The gyroscope is affected by ultrasound, which can eventually cause the drone to fall. Or in the model decision-making stage, physical attacks can be used to cause problems with the final visual recognition system of the self-driving vehicle, which may affect the vehicle’s automatic track or left and right driving or judgment of pedestrians.
With the application of AI, the abuse of technology brings out-of-control risks. The verification code verification mechanism is the first. Zhu Jifeng introduced that as early as 2017, underground black production had begun to use machine learning methods for QR code recognition training. Data shows that more than 80% of the verification accuracy can be automatically completed online registration through machine learning algorithms, which creates a series of security issues.