Credential Canaries Create Minefield for Attackers

With nearly half of all breaches involving external attackers enabled by stolen or fake credentials, security firms are pushing a high-fidelity detection mechanism for such intrusions: canary tokens.

Canary tokens, a subset of honey tokens, are manufactured access credentials, API keys, and software secrets that, when used, trigger an alert that someone is attempting to use the fake secret. Because the credentials are not real, they would never be used by legitimate workers, and so any attempt to access a resource using the canary token is a high-confidence sign of a compromise.

Last week, secrets-management firm GitGuardian released a version of the technology, ggcanary, as an open source project on GitHub. The project, tailored to Amazon Web Services (AWS) credentials, is designed to give developers a simple-to-use tool to detect attacks on their software development pipeline, says Henri Hubert, lead developer for GitGuardian’s Secrets Team.

“You can put them almost everywhere,” he says. “The best place to put them is in the CI/CD pipeline or in your artifacts used in that pipeline, such as Docker images. But you can also put them in your private repositories and your local environment. You can put them almost anywhere that is related to your developers’ work.”

As the use of cloud services and APIs have taken off, attackers have increasingly targeted such infrastructure with stolen credentials and API tokens. The average company uses more than 15,000 APIs, tripling in the past year, while malicious attacks on those APIs have jumped seven-fold, according to research released in April. In addition, nearly 50% of all breaches not otherwise due to user error or misuse make use of credentials, according to Verizon’s “2022 Data Breach Investigations Report” (DBIR). 

Tripwires Slow Down Attacks

Unsurprisingly, more companies are using canary tokens to create digital minefields for which attackers need to be wary or otherwise get caught. By seeding file servers, development servers, and personal systems with files that contain credentials or links that can act as tripwires, companies make lateral movement within their systems much more hazardous for attackers, says Haroon Meer, founder and CEO of Thinkst, a cybersecurity consultancy that created its own infrastructure for canary devices and tokens, including servers, sensors, and credentials.

Yet, attackers really have no choice: They cannot ignore potential legitimate credentials during an intrusion, he says.

“If they happen to find AWS credentials or the keys to someone’s Kubernetes cluster, attackers have to try it — it’s really hard for them to not to use those,” Meer says. “And if you get notified the moment that they try it, then you shrink the exposure window so dramatically because you are not finding out months later after they have done everything to you.”

Thinkst’s Meer likes to point to comments from penetration testers and red teams that highlight the utility of canary tokens. Attackers have to always second guess any cache of credentials, API keys, or software secrets that they find, and that slows them down, tweeted Shubham Shah, a bug hunter, penetration tester, and chief technology officer at attack-surface management startup Assetnote.

“The concept and use of canary tokens has made me very hesitant to use credentials gained during an engagement, versus finding alternative means to an end goal,” Shah said. “If the aim is to increase the time taken for attackers, canary tokens work well.”

GitGuardian’s ggcanary focuses on Amazon Web Services because of the popularity of the platform and of the infrastructure-as-code management platform, Terraform. In its best-practices document, Amazon highlights that control of AWS access keys equals control of all AWS resources.

“Anyone who has your access keys has the same level of access to your AWS resources that you do,” Amazon stated in its “Best practices for managing AWS access keys” document. “Consequently, AWS goes to significant lengths to protect your access keys, and, in keeping with our shared-responsibility model, you should as well.”

Everyone Likes Canaries

Companies such as Thinkst, GitGuardian and Microsoft are aiming to make canary tokens much easier to deploy — often in minutes.

Yet defenders are not the only ones to find uses for canaries. Attackers have also started using canary tokens as a way to detect when defenders analyze their malware.

In a recent report of an attack by the Iran-linked group MuddyWater, Cisco’s Talos Intelligence group noted the simple usage of canary tokens. The initial malware — a Visual Basic script — sends two requests for the same canary token to validate a compromise. If only a single request is detected, which would likely happen during sandboxed execution or during analysis, then the malware does not run.

“A reasonable timing check on the duration between the token requests and the request to download a payload can indicate automated analysis,” stated Cisco’s threat intelligence team in an advisory. “Automated sandboxed systems would typically execute the malicious macro generating the token requests.”

Read More HERE