Zhou Hongyi: if the digital security ability is backward, it will also be beaten

【TechWeb】7 month 27 Daily news ,360 Zhou Hongyi, the founder of the group, sent a message on his official Weibo to thank 2022 Invited by the Organizing Committee of the new wise Conference , And write an article AI weapons 、 Artificial intelligence, security, etc. have expressed their views .

Zhou Hongyi said ：

In those days, ASI · Moff is talking “ Three principles of robot ” When , There is an important principle , Robots can't hurt people . It looks like today. , This principle has been broken , In fact, all kinds of unmanned combat vehicles 、 Drones , It has been applied in the battlefield .

In the conflict between Russia and Ukraine this year , The Ukrainian army used the American spring knife UAV , This UAV has surveillance 、 Reconnaissance and strike function , stay AI blessings , A single soldier can master . Although the spring knife drone may still be remotely controlled , At the moment of the last blow, people may still give instructions . But give orders from people , To AI Automatic identification 、 Automatic judgment 、 Attack autonomously , I think it's a thin layer of window paper . Especially in AI Combined with weapons , All countries in the world are engaged in an arms race , This will also be the focus of the great powers' struggle in the future . therefore , hold AI The final decision on weapons is left to mankind , It's important to let people do the final control . I think the bottom line of artificial intelligence must be people .

In reality , Artificial intelligence security is in front of social ethics , It will also face risks and problems beyond the existing social laws and moral norms . for instance , Classic tram selection problem , And how we should determine the responsibility of driverless cars after accidents .

Self driving cars on the road , If it chooses to avoid pedestrians , You may hit an obstacle nearby , Causing casualties to the owner of the car ; If it doesn't yield , It will cause pedestrian casualties . In this case ,AI How should the program make a judgment . There is also a problem to consider , If there is an accident , Who should bear the responsibility ？ Should be AI Of ？ It's a car factory ？ It's still the driver ？ these AI Security challenges , At present, there is no clear answer .

function AI It needs calculation , Computing power requires infrastructure such as big data . But if the data center , Cloud services have been terminated , So all kinds of drones , Self driving tanks lose important brains . This scene was staged in the conflict between Russia and Ukraine .

Western countries have taken measures against Russia “ LiuDuan ”（ Broken net 、 unable to pay 、 Service interruption 、 Break certificate 、 Disconnect the domain name 、 Break public opinion ）, Including Microsoft 、IBM And Oracle, these companies stop to Russia SSL certificate , Stop domain name resolution , Stop the Internet interface , Interrupt the international transmission network , Stopped all kinds of software and services , Chip and hardware suppliers , For example, Intel cut off supply to Russia , Even Facebook 、 Twitter and others completely banned many accounts in Russia , Completely isolate Russia , Even trying to erase Russia completely in the digital world . Russia lacks the support of market-oriented digital industry , Calculate the force 、 The foundation of products and services is constrained , Completely in the situation of being beaten passively . This fully shows that backward digital security capabilities will also be beaten .

Artificial intelligence security is also an important challenge in its development , among AI The algorithm may be deceived 、 Being misled . Google once had a AI Algorithm , Identify black people as “ chimpanzee ”, Cause a lot of trouble . Now some AI vulnerability mining teams have also found ： When people walk in the street , Facing the artificial intelligence cameras that can recognize faces all over the street , Just wear special stripes or special patterns T T-shirt , The camera will think you are not alone , I won't recognize your face , So that you can realize what science fiction mentions “ stealth ” effect .

360 It has always attached great importance to the safety of artificial intelligence , And build an artificial intelligence security capability system . among ,360 The world's largest security big data accumulated by itself AI analysis , To form the 360 Cloud safe brain prototype , Provide strong data and intelligence support for industries or enterprises , Effectively analyze network threats 、 Tracing to the source ; At the same time, actively undertake “ Safe brain national new generation artificial intelligence open innovation platform ” Construction ; In addition, it also led the research and development “ Artificial intelligence security evaluation and risk monitoring and early warning platform ” Selected by MIIT “ AI industry innovation task unveiling unit ” list . For all that , Facing many AI security problems , One enterprise alone cannot solve all threats , So it is important to pay attention to ecological cooperation . We are willing to open our own secure big data , Unite the forces of industry, University and research , To build AI Safe Ecology , Focus on promoting innovation in digital security .