Security experts have been warning against the use of SHA-1 for years. The aging cryptography algorithm was an initial foray into hash-based security. The second iteration of Secure Hash Algorithm, SHA-1 was developed in 1995, and since there have been new versions, from SHA-2 to SHA-256, developed. However, SHA-1 is still in heavy use today in SSL communication, despite being warned against it for about the last six years. Last week, though, Google put the final nail in SHA-1’s coffin, delivering the first repeatable collision attack against the hash function.
On February 23, 2017, Google released its report on the first successful and repeatable collision attack on SHA-1, showcasing how the algorithm can fail “due to direct attacks on the mathematical underpinnings.” The attack was the result of two years of research on how to practically generate a collision, in collaboration with the CWI Institute in Amsterdam.
What’s a Collision Attack?
A collision attack is a basic strike against the structure of hash-based cryptography. Two distinct pieces of data are created, such two PDFs with different content, but with identical hashes. In practice, this should never occur, but there are inherent flaws in SHA-1 that allow a bad actor with serious processing power to generate such a file. The second file with the matching hash can then include malicious code to infect a system, create a backdoor and otherwise compromise it. The attack then simply has to send this infected file to the system that relies on the copied hash for its security, and it can be deceived into accepting the malicious file instead of the original.
Of course, infecting a system isn’t the only use of this attack. The same method could be used to send altered contracts, falsified medical or financial reports and more.
How Did Google Create the Attack?
Google has long warned against the use of SHA-1 and even announced that its Chrome browser would be phasing out the use of SHA-1 this year. However, in order to educate the community of the dangers of continuing to use this older algorithm, they decided to prove that this type of attack could be perpetrated.
Using the vast cloud infrastructure at its disposal, Google and researchers from the CWI Institute generated two PDF documents with noticeably different content but the same hash. The computation required 6,500 years of CPU computation to complete the first phase and another 110 years of GPU computation to complete the second phase, performing more than nine quintillion (9,223,372,036,854,775,808) SHA1 computations in total. This is, to date, one of the largest computations every performed, but as computer technology continues to evolve at breathtaking speed and IoT malware and related takeovers demonstrate hackers abilities to leverage amazingly large networks for attacks, we’re likely to see the process become much easier very quickly.
What Does This Mean for Security?
For most companies, this breach of security isn’t a big deal. Many firms have already moved away from using SHA-1 for SSL communication, embracing stronger cryptography for secure communication and file transfer. However, there are many that do still use it, creating a serious flaw in their security platforms. This report highlights how critical it is to leave antiquated forms of security behind, be it SHA-1 or passwords altogether, and embrace advanced cybersecurity technologies.