Published: August 30th, 2020
Computer scientists, among them a researcher from Los Alamos National Laboratory, have put together artificial intelligence (AI) system that pinpoints malicious codes used by cybercriminals to hijack supercomputers and mine Bitcoin and Monero.
Computers scientists have developed an artificial intelligence (AI) system that they believe will help identify malicious codes that cybercriminals use to siphon computing resources from the world’s supercomputers. The act that is also called cryptojacking is a massive challenge among individuals and corporations with enormous computer resources.
According to Gopinath Chennupati, a researcher attached to the Los Alamos National Laboratory, and working on the AI project, the system will soon stop cybercriminals and their malicious codes right on their tracks.
Gopinath said that according to the reports of recent computer break-ins in the U.S., Europe, and elsewhere, the system that his team has developed would soon prevent unscrupulous crypto miners from accessing high-performance computing facilities and taking advantage of their facilities.
The information about the system is contained in the latest edition of the IEEE Access journal. The said info said that the team’s deep learning AI model detects the abusive use of supercomputers for crypto mining purposes.
Legitimate Bitcoin and cryptocurrency miners often put together enormous computer arrays that they dedicate to discovering the digital currencies. However, unscrupulous operators have worked shortcuts that involve hijacking and taking advantage of supercomputers. Such individuals have also perfected the art of remaining anonymous and keeping their tracks and efforts hidden.
According to the team of researchers behind the new system, their program will catch illegal mining activities. The research paper explains that the AI system compares the graphs of legitimate and unlawful mining operations. Since these graphs are akin to fingerprints, the research paper adds that pinpointing illegal activity becomes easy.
It is possible to represent all programs using graphs made of nodes tied together by lines, loops and jumps. The new AI system compares the contours contained in the flow control graphs to the records on the network, just as the justice systems round up criminals by comparing the arcs and whorls on the fingertips to the governments’ databases. The graphs of programs that run on a particular computer are unique to the said machine.
However, instead of discovering a match to an established criminal program, the AI system checks to see if the graphs identify the various programs meant to run on the machine being protected.
The researchers tested the reliability of the new program by comparing a familiar, albeit benign code, to a known, unauthorized, Bitcoin mining code. The results showed that their program recognized the illicit mining operation much faster and is more reliable compared to the conventional, non-AI analyses.
The paper adds that since the system relies on an approach that compares graphs, standard techniques used by illicit cryptocurrency miners cannot escape its elaborate tentacles. Such unauthorized mining operations disguise their footprints by obfuscating comments and variables to make the graphs resemble the prints of legitimate programming.
The researchers warn that the graph system may not provide an entirely foolproof solution for the myriad of situations encompassed in the mining arena. However, they add that the solution expands the base of practical approaches that detectives in the cyberspace can use in their efforts to curtail cybercrime.
The trends noted about computer break-ins provide hope that software watchdogs such as this new AI system will soon prevent illicit cryptocurrency miners from patching into high-performance computing facilities where they loot valuable resources.
The threat of cryptojacking is worrying. According to reports, hackers successfully assembled a botnet of crypto miners that controlled more than 500,000 machines. The botnet that tech world referred to as Smominru is only comparable to Mirai botnet, the malware that almost crushed the internet. According to the cited reports, Smominru helped the illicit miners amass nearly $2.5 million worth of cryptocurrency.
Before the new AI system that the Los Alamos National Laboratory are working on, the blockchain and cryptocurrency space have tried several remedies that have not amounted to much. Among the solutions proposed include patching systems, blocking attack vectors, and monitoring abnormal CPU and GPU usage.
Jeff Edwards, a tech analyst and writer covering IT and information security, said that while these methods above are useful, technically they are only better than doing nothing. Jeff added that unless the tech world comes up with a compelling answer to cryptojacking, there exists no silver bullet that can keep crypto miners off the networks of corporations with supercomputers.
The problem with the above methods is that they are weak outside the parent system, Jeff added. For instance, patching networks is a critical step in protecting systems. However, patched systems are not safe, especially when the user visits a dubious site. The same applies in situations where a user installs a wrong app, Jeff warned.
Jeff said that the most prominent and straightforward solution is to block JavaScript from running when you are visiting suspect destinations. However, he countered the suggestion by saying that doing so is not practical in many situations involving supercomputers.
The analyst said that if anything, limiting specific capabilities of supercomputers significantly reduces their functionality, which beats the whole logic of installing them in the first place. Overall, Jeff said that cryptojacking is neither victimless nor harmless. Therefore, any efforts and steps taken to protect supercomputers are welcome.
A team of researchers, including a staffer at the Los Alamos National Laboratory, have put together an AI system that detects illicit cryptocurrency miners from patching onto supercomputers. According to industry insiders, such a system is critical now because it prevents unauthorized miners from stealing precious the supercomputers’ computing resources. The system is not foolproof. However, it so far offers the most concrete hope that stopping cryptojacking is possible. While the system is not ready for the market, the researcher believes that the situation will soon change.