CISA's first cyber decoy guide: start with one tested tripwire
CISA published its first guide dedicated to defensive cyber decoys on September 16, 2026. Here is what it says, and the smallest useful way to act on it: one realistic file, one named responder, one test that shows the alert arrives.
What CISA published
The guide covers decoys placed inside the network: tripwires, honeytokens, and decoy systems that reveal adversaries who already have a foothold. The focus is internal decoys rather than traditional internet-exposed honeypots. CISA frames the whole approach as a companion to Zero Trust: assume compromise, and add controlled, high-signal ways to detect it. CISA's words are that decoys validate Zero Trust assumptions by revealing unwanted internal activity.
The intended audience is stated plainly: critical infrastructure organizations, small to medium-sized organizations, security staff new to decoy operations or to Engage, and anyone looking for practical, low-complexity ways to improve detection and response. The prerequisites are a basic grasp of the ATT&CK matrix and the controls you already run, such as EDR, SIEM, firewalls, and identity management.
Start with a familiar workflow
The guide's own worked example is deliberately small. The security team places realistic documents with fictitious content in a project manager's personal network share. The manager knows to leave them in place. Monitoring alerts on access. Because only one person can reach the share and that person knows, any interaction is a strong indicator of unauthorized activity.
The decoy has a clear purpose, a known owner, and an observable interaction. That gives the responder a concrete starting point: who accessed it, what happened, and what surrounding activity explains it?
- 01Decoy file
- 02Monitored access
- 03Alert
- 04Named responder
Build the pilot around four decisions
Purpose. Start with an asset or workflow where better visibility would help, and write down the behavior you want to reveal. CISA's water scenario states its objective in one sentence: detect adversary movement toward OT configuration data as early as possible.
Placement. Decide where the token belongs and who owns that location. Use fictitious content and agree the placement with the owner. CISA rates honeytokens as a low-complexity starting point and lists four sourcing routes: open source, commercial, homegrown, or living off the land with the EDR, identity, and DLP tools you already have. Take whichever gets you a first token this week.
Monitoring. Decide whether you need to detect opening, copying, modifying, or an authentication attempt, and verify that your mechanism records that interaction. Our observation, not CISA's: a file that calls home when opened may stay silent when it is copied unopened.
Response. Choose an alert destination someone watches, a backup responder, and the first investigative checks. Account for approved tests, indexing, backup tools, and other routine automation when reviewing an alert.
Use the starter worksheet to record these decisions.
Test the path from interaction to acknowledgement
CISA's scenario team deploys during a maintenance window, then tests that each decoy generates its alert. Do the same, and follow the alert all the way through:
- Confirm the event was recorded.
- Confirm the alert reached its destination.
- Record the responder's acknowledgement.
- Resolve any missing step before expanding.
Keep the timestamps and repeat the test after relevant changes. A heartbeat checks liveness; this exercise checks the alert path your team will rely on. A working decoy may stay quiet for a long time; test on a schedule so the quiet is known rather than assumed.
Two details from the guide that are easy to skim past: keep detailed decoy documentation outside the production environment, restricted to the people who operate and respond to it, so a compromised network does not hand the adversary your decoy map; and define the expected false positives, IT testing being the obvious one, before the first alert rather than after.
“Cyber decoys used in a proactive cyber defense strategy help make critical infrastructure networks unfriendly places for adversaries and enhance resilience to compromise, even against living-off-the-land techniques.”
Chris Butera, Acting Executive Assistant Director for Cybersecurity, CISA, September 16, 2026
The stated outcomes are earlier detection in the intrusion lifecycle, threat intelligence gathered from intrusions and attempted intrusions, better allocation of defensive resources, and reduced mean time to detection through high-fidelity alerts. CISA describes decoy techniques as incremental, cost-effective, and scalable, deployable without major architectural change.
An interaction becomes a lead.
- Phish
- Discovery
- Steal creds
- Escalate
- Move laterally
- Exfiltrate
Configured alert path
- EventInteraction recorded
- AlertSent to monitoring
- ResponderContext to investigate
A monitored interaction gives the responder an account, an asset, and an action to investigate.
What else is in the guide: three goals, in CISA's order
The guide borrows MITRE Engage's vocabulary and is explicit about sequencing. Expose means detecting adversaries in the environment: decoys placed where legitimate users rarely or never go, configured to alert on any access, execution, or authentication. CISA's examples are fake admin-only credentials, decoy folders on executive desktops, and a fake network-scanning tool on a nontechnical user's workstation. Affect means raising the adversary's cost by diverting effort to decoys, seeding reconnaissance with a misleading picture, and encouraging exfiltration of meaningless data; the guide is careful that this buys time and detection opportunities but does not remove the adversary. Elicit means observing adversary TTPs in isolated, realistic environments, and CISA says it requires mature logging, isolation, and staff able to manage the operational and legal risk. Its recommendation: establish basic Expose capabilities first, add selected Affect activities where appropriate, and only then consider Elicit.
Honeytokens: where the guide says to start
CISA draws a clean line between honeypots and honeytokens. Its comparison table is the most quotable thing in the document.
| Attribute | Honeypot | Honeytoken |
|---|---|---|
| Level | System or network | Data or asset |
| Complexity | Medium to high | Low |
| Maturity required | High | Low |
| Primary uses | Divert adversaries; alert on access; reveal TTPs through interaction | Alert on use; show where adversaries have gained access; reveal exfiltration paths when tokens appear outside the environment |
| Types | Systems, servers, databases, web pages, clients | Files, data records, accounts, credentials, email addresses, fake URLs |
The six token types CISA names are files and documents that alert when opened, copied, or modified; fake data records placed alongside real ones; disabled or unused accounts where any authentication attempt is suspicious; fake passwords, API keys, and tokens whose use indicates compromise and helps identify where they were obtained; decoy email addresses seeded in directories; and decoy URLs with no legitimate business use. An effective tripwire, in CISA's words, is threat-informed, distinct from normal behavior, detectable in the monitoring you already run, and actionable through a written procedure.
For water utilities, follow the detection gap
CISA's notional water and wastewater scenario is the one most readers of this blog will care about. The utility builds three overlapping ATT&CK maps: where it is vulnerable, where it has detection coverage, and what its likely adversaries use. In the scenario, a ransomware group's techniques include phishing, exploitation of public-facing applications, password spraying, PowerShell, network service discovery, exfiltration, and data encrypted for impact. Overlaying the maps, the utility finds gaps around phishing beyond user reports, living off the land with PowerShell, and exfiltration targeting cloud backups. It concludes that segmentation limits direct OT exposure, so it prioritizes the IT systems holding OT configuration data and cloud backups, selects decoys through the Engage matrix (fake low-privilege accounts, decoy listening services, beacons in login pages, seeded command tokens), and validates the operation with a red-team exercise.
The useful question is the same for any organization: which important workflow needs better visibility? For OT-related placements, review permitted traffic and operational impact with the system owner. Our position for OT segments is that a decoy should listen and never scan; the guide is silent on active discovery from the decoy, and we would rather say so than let it pass.
Make the first test count
Aim to finish the pilot with an approved placement, a working trigger, a named responder, and a recorded test. Review the results together, improve what needs attention, and then expand, following the guide's own sequence from Expose toward Affect and Elicit as resources and expertise allow.
Deception Check builds decoy-based detection for water, local government, healthcare, energy, and the MSSPs that serve them, and operates an internet-facing research fleet. Our starter worksheet turns these steps into a plan your team can use, whether or not you ever become a customer.
Get the honeytoken starter worksheet. Read CISA's original guide
