This article is more than 1 year old
Boffins find that over nine out of ten 'ethical' hackers are being a bit naughty when it comes to cloud services
Then again, cloud providers aren't exactly playing the smart game either
Infosec pros and hackers regularly abuse cloud service providers to conduct reconnaissance and attacks, despite efforts by cloud providers to limit such activity.
In a recent research paper titled "Cloud as an Attack Platform" [PDF], five boffins from Texas Tech University – Moitrayee Chatterjee, Prerit Datta, Faranak Abri, Akbar Siami-Namin, and Keith Jones – describe a series of interviews they conducted with computer security pros attending the Black Hat and DEF CON conferences.
Of the 75 security professionals and hackers they spoke with as a part of a larger examination of attacker psychology, more than 93 per cent admitted to abusing cloud services to create attack environments and launch attacks.
"We observed that these professional hackers often employ common strategies to abuse the cloud platform for its resource-efficient features in order to remain stealthy and silent while probing target machines, collecting victim data, discovering vulnerabilities, and launching attacks," the paper explains.
"We did not collect any demographic data, so we can not tell apart an ethical hacker/pen-tester or malicious hackers," said Chatterjee, a doctoral candidate and corresponding author, in an email to The Register. "Moreover, at conferences like DEF CON, participants do not get a name tag, we all got a tag that says 'human.'"
Chatterjee said recruiting participants to answer questions was difficult because they were afraid of being subject to a social engineering attack. "Some of them even thought we were stealing biometric data when offered them a pen to scribble down their thoughts using some diagram," she said.
Those using cloud services for offensive operations, the researchers say, have a common pattern. They set up a Virtual Private Server (VPS) or a multi-hop Virtual Private Network (VPN) to communicate securely with virtual machines (VMs) and load them with the cybersecurity tools like NMap, Metasploit, and Wireshark so they can conduct offensive operations.
Though infrastructure-as-a-service providers try to avoid this through VM network quotas, or tools to secure accounts like AWS GuardDuty and Amazon Inspector, infosec pros can work around platform limitations.
Chatterjee said the interviewees mostly mentioned AWS, but added that Google Cloud Platform abuse has been documented too. These companies, she said, are aware that abuse happens but they don't sufficiently monitor basic accounts to stop it.
The Register asked AWS and Google for comment. A Google spokesperson pointed to the company's acceptable use policy for GCP. AWS did not respond but has a similar policy.
Staying under the radar
The boffins identified several common attack scenarios in a companion paper, "Launching Stealth Attacks using Cloud" [PDF].
Wakey-wakey! A quarter of IT pros only get 3-4 hours' kip – and you won't believe what's being touted as the 'solution'READ MORE
It outlines attack scenarios like using cloud platforms for phishing, DDoS, password cracking, rogue services, and other operations like running command-and-control servers. It also covers setting up an example attack server using Oracle VirtualBox and Kali Linux.
Both papers are scheduled to appear at the IEEE Computer Society Signature Conference on Computers, Software and Application (COMPSAC 2020) in July.
The researchers suggest some ways cloud providers might deal with abuse more effectively. One involves better customer identity verification through background checks. The availability of websites offering fake credit card numbers, the researchers say, makes it easy to create cloud accounts anonymously.
They also suggest better tracking of network usage and more intelligent VM monitoring to detect suspicious accounts. Cloud providers, they say, can enforce the use of firewalls, encourage software updates on VMs to protect against vulnerabilities, and create trusted software repositories to limit the availability of attack tools on their platforms.
Such measures, however, would impose costs on cloud providers and might alienate customers if the oversight is too heavy-handed.
Akbar Siami-Namin, associate professor of computer science at Texas Tech University and a co-author of the two papers, told The Reg there are both technical and business challenges cloud providers need to deal with.
"Security and particularly addition of security controls to infrastructure means additional costs," he said. "It also negatively impacts the 'usability' of the underlying systems. In the cloud platform, since it is a distributed environment, it is super hard to define a solid fortress in order to prevent adversaries from abusing it."
He also pointed to the business challenges entailed by adding friction to the customer acquisition process.
"Individuals are less likely to try something on the cloud, if the cloud is asking for so much private information such as credit card, name, address, etc," he said. "Therefore, the cloud providers offer some sort of basic computation power as 'free' with the hope that the potential customers (individuals) become regular customers for their cloud platforms (a long range investment)."
Those who would abuse cloud systems, he said, take advantage of this goodwill gesture.
"The cloud providers are aware of this issue," Siami-Namin said. However, from a technical point of view, it is hard to make the system bullet-proof and protected from attacks and abuse. As pointed out in our papers, perhaps they need a more mature, resilient, and monitoring tracking system empowered with AI and automated detection to identify these abuses." ®