This weekend I found an interesting article about the AI powered botnet attack:
As AI usage increasing in our daily life, there will be some negative use cases. This possibility will open a new page for the next generation cyberwar. The tools like Shodan give an ultimate data source for these attacks. The current attacks only use one and one information pieces:
A few years ago Shodan was used to find unsecured surveillance cameras, which made from a security device, a potential data leakage tool.
These days Shodan used to find vulnerable instances of MongoDB databases, which can be encrypted. This attack is more affects the companies than private persons. The size of the problem shown on the picture below:
As more and more companies using unstructured data for different analysis, which many times used for real-time decision making, this attack able to kill the whole business model for a company for weeks.
Again, these attacks are for one and one specific area, but what happens if an intelligent algorithm plan a more complex attack. It can utilize all tools, which available, to make the impact most effective. There are several factors which help the bad guys to make these successful.
Most of the companies are moving in the direction of standardization. This makes the attackers work more easy. They can use one tool for all companies where the same solution used. As we are trying to reduce the used solutions we are increasing the attack factor. Less complex passwords, factory default settings, makes the tasks more and more easy.
My opinion in the future:
The right way is a unique system design.
We can create standards, but it must be our own standards. We need to create cross validation checks, on cross platforms to ensure the whole system integrity. We need to create barriers between different components, and make them unique and secure. The best if we use custom components on both hardware and software level.
We need to analyze both the system to system, and human to system communications, to find patterns which not usual. We need to use encryption where possible, and store data multiple locations different keys. The overall key usage distribution strategy must be planned properly, to ensure it will not kill our operation, like we have seen most of the cloud failures was connected to a key change.
As I mentioned in my other articles the usage of AI and pattern analysis will be the most important in every IT. In ten years, the traditional business functionality will consume around 10 percent of the whole landscape, while the rest will be the AI and analysis layer.