Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
A canary token is a decoy credential, file, or URL that does nothing except raise an alert when something touches it. Almost no false positives, almost no cost.

What Is a Canary Token?
Most detection tries to spot malicious activity among a very large volume of legitimate activity, which is why analysts spend their days triaging alerts that turn out to be nothing. A canary token inverts the problem. It is an artefact placed where no legitimate process should ever touch it, instrumented so that any interaction generates an alert. Because nobody has a reason to open it, every alert means something. The idea is an old one, formalised by Thinkst with the free canarytokens.org service and the commercial Canary appliance, and now built into several cloud platforms.
Each token type exploits a different callback. A DNS token is a unique hostname under a domain the defender controls; resolving it tells the owner the lookup happened, along with the resolver’s address. A web bug token is a unique URL, so any HTTP request to it is an alert. A Word or PDF token embeds a reference to such a URL, so opening the document causes the application to fetch it. An AWS key token is a genuinely formatted but powerless access key pair, wired to a monitoring account so that any attempt to use it is logged.
Other common forms include Windows folder tokens using a desktop.ini directive that fires when the folder is browsed, cloned-website tokens that alert when your login page is copied onto another domain, QR codes, SVN or Git repository tokens, and Kubernetes service account tokens.
Placement is what separates a useful deployment from decorative noise. The guiding question is: what would an attacker look at next?
.aws/credentials file, in a build server’s environment, or committed to an internal repository catches anyone harvesting secrets.Payroll Export 2026.xlsx in the finance share. Nobody opens it; ransomware operators and anyone browsing for data to steal do.Cloud providers now ship equivalents. AWS GuardDuty has an extended threat detection path for decoy credentials, and Microsoft Defender for Identity supports honeytoken accounts whose authentication attempts are surfaced directly.
The appeal is the signal-to-noise ratio. A token alert needs no scoring, no correlation, and no baseline; it means a human or a process did something it should not have. This makes canary tokens one of the few detection controls that a small team can deploy in an afternoon and still benefit from two years later. They also detect categories that other controls miss, such as an insider browsing shares they have legitimate access to, or data leaving in a backup.
The limits are real. Tokens are detective only: by the time one fires, the attacker is already inside. Several types depend on an outbound callback, so a token on a network with no internet egress stays silent, and a careful attacker working in an isolated environment may never trigger one. DNS tokens reveal the resolver rather than the endpoint, which is weak attribution. And an attacker who recognises the pattern can avoid it, which is an argument for varying naming conventions and not clustering every token in one obvious folder.
Two operational points matter. Record where every token lives, because an unlabelled alert from a token nobody remembers deploying wastes an incident response. And route token alerts into your SIEM with high priority rather than into a shared mailbox, since the whole value of the control is that the alert is believed immediately.
Deception works best as a layer over sound fundamentals rather than instead of them. If you want to know whether an attacker inside your network would reach the tokens before anything else noticed, our Red Team Services test exactly that path.
A honeypot is a whole decoy system that an attacker can interact with, which takes resources to build and maintain. A canary token is a single artefact, such as a file or a credential, placed inside your real environment. Tokens are cheaper, easier to scatter widely, and generate far fewer operational demands, but they give you much less information about attacker behaviour.
Rarely, but not never. Backup software, antimalware scanners, data classification tools, and search indexers can open tokenised documents and fire an alert. Email security products that detonate attachments do the same. Exclude known scanning accounts and systems during deployment, and record the expected sources so an analyst can tell routine scanning from genuine interaction.
No. The canarytokens.org service is free and open source, and you can self-host it if you prefer alerts to stay inside your own infrastructure. Commercial deception platforms add management, reporting, and physical decoy devices. For a small team, free tokens placed thoughtfully deliver most of the detection value at no licence cost.
A skilled attacker can. Decoy credentials that never appear in logs, files with implausible metadata, and accounts that have never authenticated are all tells. Reduce the chance by using naming conventions that match your real estate, giving decoy accounts plausible attributes and login history, and spreading tokens rather than grouping them in a single obvious location.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.