In Plain Sight: How Hackers Exfiltrate Corporate Data Using Video

By Kaushik Narayan, Chief Technology Officer, Skyhigh Networks

Consumers and companies are embracing cloud services because they offer capabilities simply not available with traditional software. Cyber criminals are also beginning to use the cloud because it offers scalability and speed for delivering malware, such as in the recent case of Dyre, which used file sharing services to infect users. The latest evolution of this trend is attackers using the cloud to overcome a key technical challenge – extracting data from a company. Under the cover of popular consumer cloud services, attackers are withdrawing data from the largest companies in ways that even sophisticated intrusion prevention systems cannot detect.

Previously, researchers at Skyhigh uncovered malware using Twitter to exfiltrate data 140 characters at a time. Skyhigh recently identified a new type of attack that packages data into videos hosted on popular video sharing sites, a technique difficult to distinguish from normal user activity.

The Industrialization of Hacking

The target of these attacks ranges from customer data such as credit card numbers and social security numbers to intellectual property, which can include design diagrams and source code. In recent years, hacking has undergone a revolution. Once a hobbyist pursuit, hacking is now performed at industrial-scale with well-funded teams backed by cartels and national governments. Stealing data is big business, whether to compromise payment credentials and resell them for profit or to gain access to intellectual property that could allow a competitor to catch up on years (or decades) of research and development.

In response, companies have made significant investments in software that can detect telltale signals that attackers have gained access to their network and are attempting to extract sensitive data. With these intrusion prevention systems in place, it can be quite challenging for attackers to remove a large amount of data without being discovered. In the same way that thieves would find it difficult to sneak bags of money out the front door of a bank undetected by guards and security cameras, today’s cyber criminals need a way to mask their exit. That’s why they’ve turned to cloud services to make large data transfers.

Their latest technique involves consumer video sites. There are two attributes that make video sites an excellent way to steal data. First, they’re widely allowed by companies and used by employees. There are many legitimate uses of these sites such as employee training videos, product demos, and marketing the company’s products and services. Second, videos are large files. When attackers need to extract large volumes of data, video file formats offer a way to mask data without arousing suspicions about a transfer outside the company.

How the Attack Works

Once attackers gain access to sensitive data in the company, they split the data into compressed files of identical sizes, similar to how the RAR archive format transforms a single large archive into several smaller segments. Next, they encrypt this data and wrap each compressed file with a video file. In doing so, they make the original data unreadable and further obscure it by hiding it inside a file format that typically has large file sizes. This technique is sophisticated; the video files containing stolen data will play normally.

They upload the videos containing stolen data to a consumer video sharing site. While they’re large files, it’s not unusual for users to upload video files to these types of sites. If anyone checked, the videos would play normally on the site as well.

After the videos are on the site, the attacker downloads the videos and performs the reverse operation, unpacking the data from the videos and reassembling it to arrive at the original dataset containing whatever sensitive data they sought to steal.

What Companies Can Do to Protect Themselves

Traditional intrusion detection technology generally does not detect data exfiltration using this technique. One way to identify this attack is an anomalous upload of several video files with identical file sizes. To identify this type of activity, what is needed is a big data approach to analyzing the routine usage of cloud services in the enterprise to detect these anomalous events.

Skyhigh analyzes all cloud activity to develop behavioral baselines using time series analysis and machine learning, and identified the attack in the wild at a customer site. Importantly, the detection relied on analysis of normal usage activity rather than detecting malware signatures that don’t exist before the attack has been catalogued. Skyhigh’s approach requires no knowledge of the attack before it’s detected.

Companies can proactively take steps to protect themselves by limiting uploads to video sharing sites while allowing the viewing or download of videos. Deploying a cloud-aware anomaly detection solution can also give early warning to an attack in progress and either block it from occurring or quickly allow a company to take action to stop the attack and prevent additional data from being exfiltrated.

The volume and sophistication of attacks is increasing. In this threat environment, companies must take additional steps to protect data while allowing the use of cloud services that also drive innovation and growth in their businesses. State-sponsored attacks and sophisticated criminal organizations are now using the cloud as a delivery vehicle for malware and as an exfiltration vector, but companies can also take advantage of a new generation of cloud-based detection and protection services to safeguard their data and protect themselves. Download our cheat sheet to learn other actionable steps for reducing risk to data in the cloud.