In a threat bulletin published on our blog in December, we explored the details of the major breach at Sony Pictures Entertainment orchestrated by the “Guardians of Peace” (also known as #GOP). The attack resulted in the release of much more than gigabytes of valuable data, including dozens of digital certificates and SSH and SSL private keys—keys that could allow privileged-user access to the entire internal network of Sony. Once on the network, using these compromised keys, the bad guys likely remained undetected for weeks, months, or even years and had unfettered access to systems and data. And now that these private keys are in the wild, more bad guys could further infiltrate Sony.
Since the news initially broke there have been multiple updates and discoveries, and I suspect there will continue to be more. This is a huge, complex breach that would have been very difficult to stop—but within it are a few important lessons for other enterprises to take to heart.
The threatscape has changed. Cybercriminals are (and have been) looking to compromise cryptographic keys and certificates, and this Sony breach is just the latest in a series of several incidents using the same exploit. Looking back to April 2011, Sony’s PlayStation Network (PSN) suffered a significant breach that exposed names, addresses, and credit card data belonging to 77 million user accounts and shut down the PSN for several weeks. The breadth of the data exposed in that attack indicated that data which should have been encrypted was not.
Many believe attackers obtain keys to allow data or transmissions to be decrypted, but they do more than that. We’re now seeing again that bad guys gain access to private keys, allowing access to a treasure trove of sensitive internal data such as payroll and financial management, which was the case for Sony. And because one key in this breach was for Audible Magic, an entertainment service that identifies stolen digital media, this could have been one of the ways the to-be-released movies were accessed.
Incident response must involve replacing all key and certificates. The incidents at Sony should sound familiar: we’ve seen cybercriminals from Mask, Crouching Yeti, APT18 and others misuse SSL certificates and SSH keys. In these cases and others, attackers can gain unauthorized access to a system with elevated privileges using a compromise certificate or SSH key (like Edward Snowden), expand their attack by gaining more data or misusing a compromised system, gain access to continually more systems, and leave behind backdoors as we’ve seen with Shellshock.
The only way to remediate this is to change out all keys and certificates. Otherwise, bad guys retain the presence and p0wning of networks. Advice from Erik Heidt at Gartner on responding to incidents like Heartbleed provides a good template: new keys must be generated, new certificates issued, old certificates revoked, and the replacement of new keys and certificates validated. Getting back to a known, good state can’t mean relying on the same keys and certificates that are increasingly being misused.
So why hasn’t Sony simply replaced these keys yet? Well, that’s much harder to do than it sounds. The first problem is that most enterprises aren’t aware of all of the keys and certificates they have, where and how they are used, and who is responsible for them—from SSL and SSH to code signing, VPN, WIFI and more. Most organizations use many Certificate Authorities (CAs) and there are increasingly more applications, devices, and cloud services that need to use keys and certificates.
The second issue is that many security teams don't know how to detect which keys and certificates are being misused. The Ponemon Institute found that in the average Global 2000 organization there were on average more than 17,000 SSL keys and certificates, including those from internal CAs and self-signed.
Finally, security teams don’t have the means to automate remediation. Security and response teams haven’t been tooled to generate new keys, issue new certificates, and revoke old ones—just look at the poor level of remediation from Heartbleed. Venafi research from 2014 found that Heartbleed remediation for 97% of the vulnerable G2000 SSL certificates had not been completed. In addition, University of Maryland research published in November 2014 validated this widespread non-remediation. At any one time, enterprises need to know all of the keys and certificates that are in use and then be able to respond quickly to replace and revoke them when needed.
This is another clear example proving keys and certificates must be secured and protected. Here’s a case of history doomed to repeat itself as long as the same attack pattern continues to work (and it does): get the keys and own the kingdom. As recent breaches have proved—and as the Cost of Failed Trust research revealed almost two years ago—all it takes is one compromised key or vulnerable certificate to cause millions of dollars in damages. Failing to continuously surveil all keys and certificates, enforce a policy, detect misuse and anomalies, and respond and remediate by replacing them with good keys and certificates means that security will continue to be undermined. By misusing keys and certificates bad actors can undermine and circumvent many of the most critical security controls, including strong authentication, DLP, sandboxing and privileged access management.
So while many IT security pros and incident response teams continue to focus on who was behind the Sony breach, what their intention was, and what data was stolen or exposed, let’s take this opportunity to learn an important lesson. We should start 2015 by working to better secure and protect SSL keys and certificates, SSH keys, and the range of keys and certificates increasingly being used for VPN, WIFI, and MDMs.