你所在的位置: 首页 > 正文

人工智能如何影响网络安全

2019-07-22 点击:1640

人工智能是一门新的技术科学,研究和开发用于模拟,扩展和扩展人类智能的理论,方法,技术和应用。它由不同的领域组成,如机器学习,计算机视觉等。自人工智能诞生以来,理论和技术日趋成熟,应用领域不断扩大。可以想象,未来人工智能带来的技术产品将成为人类智慧的“容器”。如今,它已经开始关注长期的网络攻击,人工智能将重塑网络安全的未来。

a2657a9269114167bfad0d822b6a1488

我们现在在一个紧密联系的地球村。许多从小就生活在数字时代的人甚至不记得苹果手机出现之前的时代。随着智能家居的兴起,我们越来越多地将照明,门锁,摄像头,恒温器甚至烤面包机连接到家庭网络。通过移动应用程序或语音自动管理我们的家庭,显示了我们在过去几年中取得了多大进步。但是,在我们追求“酷”和“方便”的过程中,很多人都没有停止考虑自己的网络安全责任。

今天的网络安全风险远远高于大公司的数据泄露风险,连接到网络的所有内容都是攻击的目标。尽管数十亿美元投入了网络安全,但报告的网络攻击数量和入侵规模仍在不断增加,复杂和破坏性的网络攻击对多个行业的复杂性和规模日益增加。特别是,我们的关键基础设施领域存在漏洞,例如能源,核能,水,航空和关键制造业,这使得它们成为网络犯罪攻击的目标,甚至落后于某些国家资助的网络。攻击目标。

618196579bef4f67965bfcb78bfb23de

不幸的是,90%的网络攻击使用人为错误或人们不作为是入侵的主要原因。已有无数的例子,例如以低于8美元的价格出售的DNA数据库,以及美国政府的黑客攻击导致560万联邦雇员的指纹泄露。事情已发展到无人能够预测将来会发生什么的程度。毕竟,学习在线交易技巧从未如此简单:机器学习软件随时可用,视频教程只是一种搜索。通过自动编辑潜在受害者的内容信息,网络犯罪分子可以迅速对企业或个人造成严重损害。人们呼吁并迫切需要一种方法来完全保护我们的网络安全。

幸运的是,新兴的人工智能机器学习模型给我们带来了希望。它采用主动方法,而不是传统的被动响应,为我们提供更好的保护,以抵御这些复杂的威胁。实质上,最重要的变化是在攻击发生之前阻止攻击。在这些前沿领域,人工智能的预测能力和机器学习演化能力的使用可以为安全系统供应商和包括个人和企业在内的所有人提供优势:

思科系统公司(思科)预计,到2020年,全球连接设备的数量将从目前的150亿增加到500亿。由于硬件和软件资源有限,这些设备中的很大一部分没有基本的安全措施。最近针对Curbs On Security的大规模拒绝服务攻击生动地证明了被黑客入侵的物联网设备的强大功能。

Even more frightening is that the malware source code used to launch the attack is quickly released to the public and can now be used to attack any business or individual. Internet of Things security is one of the most prominent areas of artificial intelligence technology. A lightweight predictive model based on artificial intelligence that detects and blocks suspicious activity on the device or on the network in real time, even on low-computing devices.

0074df77935f42ad956ebfc5265c3053

File-based attacks remain one of the main cyber attack vectors. The most common file types used for cyber attacks are executable files (.exe), PDF files, and MS Office files.

With a small change in one line of code, you can generate new malicious files with the same malicious purpose but different signatures. Its small changes in behavior deceive anti-virus programs based on legacy signatures, as well as more advanced heuristic-based advanced endpoint detection and response (EDR) methods, and even network-level solutions. Methods such as sandbox technology.

A key issue for security teams is the excessive number of security alerts received each day that can cause alarm fatigue. Companies in North America process an average of 10,000 security alerts per day! In many cases, alarm fatigue can cause a malicious attack signal to escape the radar range, even though it has been marked as a suspicious signal. This requires automatic integration of events by integrating internal logging and monitoring systems with external threat intelligence services by running advanced associations between multiple sources. This cutting-edge technology for cyber defense is very popular because it solves the problem of large enterprises operating their own security operations center (SOC).

xxQuantifying an organization's cyber risk is challenging, mainly because of the lack of historical data and the need to consider a large number of variables. Organizations interested in quantifying risk (and third parties evaluating these organizations, such as cyber insurance companies) must go through a lengthy tedious network risk assessment process, primarily based on questionnaires to see if there are any cybersecurity standards that meet the available cybersecurity standards. Qualitative measures, as well as an organization's management and risk culture. This method is far from enough to truly reflect the current serious situation of cyber risks. Artificial intelligence technology can process millions of data points and generate forecasts, which may be a successful way for organizations and network insurers to obtain the most accurate network risk estimates.

337621d7a60249149f8e19e390512373

Because each organization has its own unique traffic behavior, detecting unusual traffic that may represent malicious activity is a huge challenge. To find the association between protocols without relying on intrusive deep packet inspection, you need to analyze thousands of associations between the myriad of metadata from internal and external network traffic.

According to Ericsson, there are more than 2.5 billion smartphones worldwide and it is expected to reach 6 billion by 2020. According to research by the application security company "Arxan", 56% of iOS apps and 100% of Android apps have been blacked out in the top 100 apps on iOS and Android. The two apps available in the leading app stores "Google Play" and the Apple App Store have broken the 2 million mark, highlighting the need for highly accurate and automated categorization of mobile apps. This classification method must be sensitive to the most subtle confusing techniques to distinguish between malicious and green applications. This task can be delivered to an advanced, cutting-edge classification capability artificial intelligence technology.

c27019429a234223b5cfd3483436247f

xx人工智能和机器学习具有如此强大的功能,并不意味着我们可以高枕无忧了。正当企业和政府当局开始了解人工智能和机器学习在保护他们方面将发挥的作用时,犯罪分子也在使用同样的工具来绕过防御。模仿人类行为并试图战胜防御的人工智能攻击,将是好人与坏人之间人工智能之战的开始。为了提供足够的保护,机器学习模型对威胁的检测和反应必须更快。技术的进步使得安全系统的崛起成为可能,这些系统总是在学习,适应和寻找新的方法来快人一步地掌握那些现在没人掌握的攻击手段。

人工智能的攻防之战不断展开,在网络上关于它的负面评论也渐渐浮现。有些人们抨击人工智能的“黑暗面”,认为没有人工智能就不会有网络攻击。尽管人们很容易将威胁的规模归咎于科技发展,但我们要记住,人工智能只能执行其人类主人为其编写的程序。所以,所谓的人工智能的“黑暗面”,只是人性中最坏方面的反映而已。不过毋庸置疑的是,我们正在进入一个新的数字时代,人工智能和机器学习无疑将重塑网络安全的未来图景。

XX

日期归档
中江门户网 版权所有© www.hx-wy.com 技术支持:中江门户网 | 网站地图