Machine learning and artificial intelligence are transforming from a proof-of-concept process to a functional enterprise infrastructure. As financial support for these technologies continues to rise, it is foreseeable that the popularization of artificial intelligence is inevitable.
Machine learning (ML) and artificial intelligence (AI) are transforming from a proof-of-concept process to a functional enterprise infrastructure. As financial support for these technologies continues to rise, it is foreseeable that the popularization of artificial intelligence is inevitable.
But the application of digital intelligence brings new risks: IT experts are somewhat at a loss in the face of this drastic change, and cybercriminals are trying to invade new tools.
The current security team is overworked and understaffed, but there are still people who worry that AI tools will eventually replace human expertise.
In response to these concerns, the Phys.org website pointed out in June 2018 that discussions about artificial intelligence and automation are now dominated by two kinds of ideas. One is disaster predictors, who are afraid that robots will replace human jobs, and the other is The optimists, they are dissatisfied with the new technology and think that there is nothing new in the world.
However, research shows that these technologies are only suitable for replacing certain specific tasks, rather than completely eliminating certain occupations. The Verge reported in June 2018 that an experimental program of the US military will use machine learning to better predict when vehicles need repairs, which not only reduces costs, but also reduces the pressure on technicians.
The same is true for IT security, using smart tools to complete the heavy data maintenance and data collection work, thereby freeing up technical experts to complete other tasks.
Will machine learning reduce or increase insider threats?Although new technologies will not replace human jobs, they will increase internal threats. All companies have internal threats, such as malicious acts of deliberately stealing data or unintentionally leaking company information. AI and machine learning do not have human thoughts and characteristics to cause the risk of these data leaks. It stands to reason that they should create a more secure environment.
However, this is not the case.
As CSO Online pointed out in January 2018, malicious attackers will also use this technique to contaminate data pools, thereby causing internal threats. By tampering with the data input, the attacker can also destroy the data output until it is too late for the company to find out.
At the same time, according to a report by Medium in May 2018, a more subtle attack is also growing, namely adversarial samples. By creating fake samples of the boundaries of AI decision-making capabilities, cybercriminals can force misclassifications, thereby destroying the basic trust of machine learning models.
How to defend against internal threats caused by AI?With the increasing application of smart tools, how can companies defend against more serious internal threats?
It is best to do the following:
• Establish interpersonal relationships: These new tools are only effective when performing specific tasks. Therefore, companies need to appoint personnel to study and research new tools. Once they know the tools well, they can establish a solid line of defense against potential threats.
• Develop a check and balance system: Does the reported data match the observations? Has it been independently verified? As more key decisions are transferred to artificial intelligence and automation, companies must develop inspection and balancing systems that compare output results with trusted benchmark data.
• Purposeful deployment tools: In many ways, the rise of smart technology reflects the rise of cloud technology. At first, as a new technology, this solution quickly became a necessary condition for digital transformation. At this time, there may be a similar "more is better" trend, but this ignores the key role of artificial intelligence and machine learning as a method to solve specific pain points. We should not blindly follow the trend and introduce new technologies. Instead, we should start from discovering small data-driven problems and install smart tools to solve this problem. Establish a zero-trust model of data access for smart tools: When the attack surface is essentially restricted, it is easier for us to suppress potential threats.
Machine learning and artificial intelligence tools are gaining corporate support, and fortunately, they are unlikely to replace IT technicians. Looking to the future, although invaded intelligent tools and adversarial artificial intelligence pose potential threats to the next generation, human assistance can actively respond to these threats.
Big Size Switch And Socket,Big Size Switch,Big Size Socket,Big Size Switch Socket For Sale
ZHEJIANG HUAYAN ELECTRIC CO.,LTD , https://www.huayanelectric.com