Cyber-forecast no. 1: Artificial-intelligence-Fuzzing
A Trend in the world of Cyber-Criminals is called Fuzzing. the Fuzzing is a highly developed technology in laboratory environments to detect vulnerabilities in Hardware and software interfaces and applications. Invalid, unusual, or semi-random data are introduced in an interface or program. In the connection it is checked whether events such as crashes or memory leaks may occur.
There are only a handful of professionals who have the necessary Know-how to develop effective Fuzzing Tools. Their use in Cyber-Criminal is usually limited to simple applications such as Distributed-Denial-of-Service (DDoS) attacks. Zero-Day attacks, i.e., attacks that take place on the same day discovered this exploited vulnerabilities in the Software, however, are still rare. With artificial intelligence (AI) will change probably.
Christian Vogt About the experts
Christian Vogt is a Senior Regional Director Germany at Fortinet, a leading provider of comprehensive, integrated and automated Cyber Security solutions. Fortinet equips its customers with intelligent, gap loose protective measures against the rapidly growing Cyber-threat environment.
Zero-Day attacks. automated
by Using AI and machine learning models can also develop Cyber-Criminals Fuzzing programs and training So you are able to Zero-Day vulnerabilities, automates, and accelerates to uncover. By a trained attacker routed, could this supervised machine learning repeat the approach continuously. This also means that a Cyber-Gangster can perform a combination of attacks to Zero-Day detect vulnerabilities continuously and to take advantage of. In an environment of potentially infinite Zero-Day attacks are possible, would be self-advanced IT Security Tools like Sandboxing quickly.
Zero-Day-Mining-as-a-Service, i.e. the individual development of Zero-Day Exploits, will change the handling of company Security. Because there is no way to predict where these Zero-Day Exploits are, how you can defend yourself adequately against them. Particularly difficult in the fight against automated, AI-driven Cyber-attacks have it isolated, older Security Tools that many companies use today in their networks.
Cyber-forecast no. 2: swarm-based Botnets, “as-a-Service”
in the light of rapid progress in the field of swarm-based intelligence swarms in the Cyber-use sphere, in the meantime, both the attack as well as defense. It was recently developed by scientists in Hong Kong, a new methodology based on the natural swarm behavior uses for the control of nano robotic clusters.
Using this technology, large flocks can be intelligent Bots to create work both collaboratively as well as autonomously. This is how Zero-Day Mining also impact on the business models of Cyber-Criminals.
Still, the criminal Ecosystem is very strongly tied to the human factor: Professional hackers to create custom Exploits for a fee. Even new developments, such as Ransomware-as-a-Service require that so-called Black-hat engineers, a number of resources to use, exploit, develop, validate, and back-end C2-manage Server. But that will change With the business model of Autonomous, self-learning swarms “as-a-Service” reduces the amount of direct interaction between the Hack buyers and the hackers drastically.
“What’ll it be?” À-la-Carte Swarms
The buyer of such a criminal program may choose different types of swarms for a user-defined attack, almost “À la Carte”: built-in swarms, the use of machine Learning to a device or network; swarms that perform KI-Fuzzing, Exploit-detect points; shoals, which move laterally through a network, to expand the surface area of attack; swarms that escape detection, or specific data gathering goals, and exfiltrate or swarms, crossing the border between the physical device and Cyber-space, to take control of the physical and networked resources on a target.
Cyber forecast no. 3: Machine Learning is captured
Machine Learning is already used in many of the IT Security. Such devices and systems can be trained to perform certain tasks autonomously, such as countermeasures against a detected attack. Machine Learning can also be used to detect behavior in the network and to identify, on the basis of behavioral analysis of complex threats. In addition, tedious manual tasks can also be easily used in a trained System – such as the Monitoring of the equipment as a result of their exposure to current risk trends, and the automatic installation of Patches or Updates.
But where there is much light there is also shade. Instead of trying to outwit a machine Learning-improved System or to outdo, Cyber-Criminals are learning to manipulate the machine process itself. The methodology and the tools to train a device or System for a particular task are heel, at the same time his Achilles. If an attacker, for example, manages to compromise a machine learning system, could he not exercise devices or systems so that you install Patches or Updates. So the Computer would be vulnerable to attack. He could manipulate a System so that it ignores certain types of applications or commands, or the traffic will not be logged, to leave no traces.
to prevent this, you need to pay attention to Security-Responsible to ensure that all resources and protocols of machine learning are carefully monitored and protected.
This is an excerpt from the predictions about the Cyber-threats 2019 from FortiGuard Labs < / strong> . read part 2 of the series (coming soon), with what defense strategies of the company and the perfidious Tricks of the Cyber-Criminal can contain.
FOCUS Online/Glomex If you type in certain words in Google, you receive uncanny messages Cybercrime Artificial Intelligence