In brief
- Los Alamos researchers have found a way to stop crypto miners infecting supercomputers.
- Crypto miners use computing power to secretly extract crypto, such as Bitcoin.
- The Los Alamos researchers claim their AI-based solution is superior to more conventional protection.
A group of researchers at Los Alamos, the US science laboratory that created the atomic bomb, have created an artificial intelligence system that keeps bitcoin miners out of supercomputers.
Occasionally, hackers manage to infect supercomputers—some of the most powerful computers in the world, with processing speeds hundreds of thousands of times faster than even the top gaming PCs—with cryptocurrency minersminers.
The idea is to suck up all that power and use it to mine cryptocurrencies, such as BitcoinBitcoin. They’re the scourge of supercomputer operators, who preside over computational power usually reserved for top scientific researchers. In May, for instance, at least a dozen European supercomputers had to shut down due to crypto mining attacks, according to the BBC.
Those affected, or potentially at risk, might then appreciate the work of the computer scientists at Los Alamos National Laboratory, who have devised an artificial intelligence system to stop these attackers in their tracks.
Attention AWS customers: You might be running a malicious crypto miner
Editor's note: This article was updated to show that the hacker made the malicious file available as a shared file, and that it was not available through AWS Marketplace. Mitiga, an Israeli cybersecurity firm, is advising all customers of Amazon Web Services running certain types of programs to check that they haven’t been infected by a malicious Monero crypto miner. In an advisory note today, Migita said that anyone running EC2 instances based on Community AMIs (Amazon Machine Images) are vulne...
They explained it all in their paper, “Code Characterization With Graph Convolutions and Capsule Networks,” which was published by the IEEE, one of the top bodies in Internet infrastructure.
Whereas other systems try to search for malicious code, the Los Alamos researchers’ AI system makes sure that the supercomputer runs only programs that are supposed to be running on its hardware.
According to a blog post, “instead of finding a match to a known criminal program, however, the system checks to determine whether a graph is among those that identify programs that are supposed to be running on the system.”
The researchers ran some tests and found that it finds malicious crypto miners “much quicker and more reliably than conventional, non-AI analyses.”
