Blockchain and artificial intelligence (AI) are two industries that often exhibit similar characteristics and draw interest from like-minded individuals. Blockchain startup SingularityNET has even created an open marketplace for developers and organizations to trade AI algorithms with each other. Now, as the two technologies evolve together, some of the more impressive benefits of using AI for security in the cryptocurrency sector are beginning to reveal themselves.
One of the most significant flaws in blockchain security is the way in which networks can be manipulated using massive amounts of computing power. Many smaller blockchain networks with limited miners and a low hashrate are at constant threat of a ‘51% attack‘ – when a group of miners can take control of the network by majority computing power. Since the first cryptocurrency, Bitcoin, was invented over a decade ago, there have been several high-profile thefts that have seen billions of dollars worth of cryptocurrency stolen from online wallets and exchanges. These thefts are often highlighted in the media and cited as one of the main hurdles that cryptocurrency adoption faces, as new customers are understandably reluctant to use the services.
How can artificial intelligence help?
Recently, several projects have begun investigating how artificial intelligence can be applied within cryptocurrency networks to recognize malicious behavior on a blockchain and counteract it. Not only does this remove the possibility of human error, but the self-learning nature of AI means the system is continuously improving its defenses.
Manipulation of blockchain networks doesn’t just come in the form of hacks or thefts. Due to the fact that crypto miners only profit when selected to verify a blockchain transaction, it becomes attractive for them to try and alter the process and increase their chances. A new crypto project named Velas has developed a consensus method to combat this, called AI-enhanced Delegated Proof of Stake (AI-DPoS).
Until now, it’s been difficult for blockchain networks to process high-volume transactions securely without risking a decrease in the decentralized nature of the system. The Velas AI-DPoS consensus method uses the speed and automation of AI to reward users for exhibiting good behavior and correctly using the network. No matter how much computer power a hacker as at their disposal, they would be unable to manipulate the system as the AI would immediately recognize the non-conformist behavior.
Sybil attack protection
Artificial intelligence also promises to help protect against so-called ‘Sybil’ attacks – where a group of hackers creates several hundred false identities for the purpose of inflicting majority influence over a network. Human beings would not necessarily be able to recognize the deception quick enough, but AI would instantly pick up on subtleties that are unavoidable in such an attack.
Since the majority of hacking methods rely on computer-generated algorithms to develop an attack, a system that thinks like a computer is required to defend against them. AI defense systems are continually evolving at an exponential rate, ensuring that they are always one step ahead of the hackers.