May and June 2017 saw the outbreak and rapid spread of WannaCry and NotPetya across the world. Though the initial infection vectors differed, both of these worms leveraged the same Server Message Block (SMB) vulnerabilities for lateral propagation and privilege escalation, though NotPetya added a couple of extra tricks to its bag.
These SMB vulnerabilities – EternalBlue and DoublePulsar – stemmed from a leak of NSA-authored hacking tools released by The Shadow Brokers. In both cases, the malware delivered was overt in nature, contributing to fast detection times and, in the case of WannaCry, the rapid discovery of a kill switch which was used to halt the attack.
When The Shadow Brokers dumped the cache of tools onto the Internet, Rapid7 reported that security researchers went from feeling "like kids in a candy store" to being disinterested as they realized that "the exploits were antiques and had all been patched." However, as time and ransomware actors would go on to prove, "even though we thought we were safe against these non-zero-day, unexciting attacks, we were not." And although vulnerable servers should not have been "exposed to the public Internet in an unrestricted manner," over 250,000 machines were infected by WannaCry within the first day. This was also not the first time that a cryptoworm had leveraged vulnerabilities that had been patched years earlier by the vendor.
As the WannaCry and NotPetya attacks progressed, we saw reports of breaches from the NHS, telecommunications service providers, critical infrastructure providers, vehicle manufacturers, airports and logistics companies, and even speed camera operators. But for each of these thousands of companies, across many industry verticals, the impact could have been much worse, if the payload had have been different. What if it had targeted and exfiltrated NHS patient records? What if it had modified shipping or customer manifests? What if it had disabled speed cameras or worse, moved laterally and modified traffic light sequences? What if the attack was more covert in nature? Would we have ever known?
Over the last six years, Mandiant analysts have reported a reduction in the median breach detection time from 416 days (2012) to 99 days (2017). And while, on the surface, this looks positive, it worryingly corresponds to an increase in the percentage of breaches reported by internal sources from 6% (2012) to 47% (2017), during the period in which we have seen a massive boom in ransomware innovation and activity.
So, I wonder, if ransomware attacks are leading to an increase in the percentage of internal breach notifications, and driving the median breach detection time down, thanks to their sheer volume and overt nature, how long are the covert attacks going undetected, before ransomware actors start leveraging their Tactics, Techniques and Procedures (TTPs), alerting us to the failings of our security architectures and policies, forcing us to make a change?
Until we see broader adoption of machine learning for discovering new threats, more automated sharing of threat intelligence between security vendors and security products, and the ability to leverage the network to shut down attacks at the source, we have to ask ourselves – is ransomware the tripflare in the modern cyberwar that we can’t afford not to have?