ABRACADABRA Labs Resource directory

Topic · 83 links

Cyber deception

Honeypots, honeytokens, and decoys: whether they change attacker behavior, how to use them well, and how they hold up against AI-driven attackers.

Deception has moved from a niche practice to mainstream guidance. CISA now recommends decoys alongside zero trust, and the UK's NCSC has trialed commercial products across 121 organizations. The case practitioners make is practical: attackers increasingly log in with stolen credentials and blend in with normal admin activity, and a fake credential or file that nobody legitimate should touch gives one of the few alerts that is almost never a false alarm. The newer question is what happens when the attacker is an AI agent. Early studies suggest models take the bait more readily than people do, and the same models are being used to build more convincing decoys. The evidence is thinner than the enthusiasm. Many claims that deception works trace back to one set of experiments with human red teamers, and the annotations say so where it applies. Real-world deployments raise problems the lab work didn't, including poor configuration, noisy alerts, and inconsistent vocabulary across vendors. There's also a live disagreement about secrecy. Most organizations won't say they use deception, but research suggests that attackers who know deception is present lose confidence and slow down.

Start here

Six links, best read in order. Cyber deception means setting traps for hackers: fake files, passwords or computers that raise an alarm when touched. You'll see evidence that these traps work, learn the basic ideas, see a free tool in action, dig deeper into the research, and finish with how attackers spot traps. the full set of paths is on Paths.

  1. Imposing a Cyber Penalty Against Attackers with Cyber Deception

    Start with the evidence, not the theory: a short practitioner summary of the one experiment everyone cites.

    Readable summary of the Tularosa findings for practitioners: 52% of attacker commands targeted decoys, and exploit failures doubled under deception. Read this before the two formal papers.

  2. Deception Techniques in Computer Security: A Research Perspective

    Now get the vocabulary. Four axes that let you place any technique you meet later.

    Classifies deception along four orthogonal axes: goal, unit, layer, and deployment mode. The cleanest taxonomy in the literature; the ACM version is paywalled, so the authors' copy is linked first.

  3. MITRE Engage

    The vocabulary again, but in the form organisations and vendors actually use to talk to each other.

    MITRE's adversary engagement framework: prepare, operate and understand phases over a goal/approach/activity matrix, with a starter kit. Successor to MITRE Shield, and the common vocabulary most deception programmes end up using.

  4. Canarytokens

    Stop reading and plant something. Fifteen minutes to a working detection with no infrastructure.

    Hosted honeytoken generator, free: files, URLs, credentials, cloud keys that alert when touched.

  5. Examining the Efficacy of Decoy-based and Psychological Cyber Deception

    Back to the evidence, this time the full experimental design and what the four conditions actually showed.

    The results paper from the Tularosa data: a four-condition design finding that decoys plus telling attackers deception is present changes behaviour most. The strongest peer-reviewed evidence that deception imposes real cost.

  6. A Review of Honeypots: Fingerprinting Techniques, Detection, and Evasion Mechanisms

    Finish with the other side: how attackers spot decoys, so you do not over-trust what you just deployed.

    The counter-deception side: how attackers fingerprint honeypots through behavioural and protocol tells, and how weak existing anti-fingerprinting defenses are. The corrective to optimistic deployment literature.

Everything in this topic

Grouped by subtopic, then by source type within each subtopic.

Threat reporting 1

Vendor and government reporting on observed misuse of AI systems.

  • Amazon Threat Intelligence observes threat actors running cybercrime campaigns with ARTEX AI agents

    Amazon's MadPot decoys posed as LLM inference APIs and, from 3 October to 5 November 2026, caught requests from ARTEX agents carrying their own campaign memory. Two campaigns are described, one stealing AI provider keys from exposed gateways to fund more hunting and one building deepfakes to pass exchange identity checks, but no counts of requests, sensors, keys or targets are given. Every outcome is the agents' own claim, which Amazon did not observe and cannot verify. Vendor threat report with indicators.

    vendor-authored

Honeypots and honeynets 15

Decoy systems built to be attacked and observed.

  • A Hybrid-Simulation LoRaWAN Honeypot for Detecting RF and Protocol Attacks Against IoT Networks: A Cyber-Electromagnetic Activities Perspective

    Hybrid-simulation honeypot for LoRaWAN that instruments the radio side as well as the protocol layer, aimed at RF and protocol attacks on low-power wide-area IoT networks. Whether evaluation used live RF traffic or simulation alone is the point to check before citing it. Springer CCIS chapter;. Read after guan-2025-iot-honeypots.

    paywalled
  • Multi-service honeypot framework for high-fidelity cyber-attack data collection and machine learning-based threat detection in cloud environments

    Cloud-hosted honeypot framework running several emulated services to collect attack traffic, then training machine-learning models to classify what it captured. Accuracy on traffic the same honeypot collected says little about performance elsewhere, and whether the captured data is released is unconfirmed. Scientific Reports, open access.

  • Cyber-Physical Deception Through Coordinated IoT Honeypots

    Coordinates multiple IoT honeypots so their simulated physical and network dependencies stay consistent under multi-source probing, closing a detection channel single decoys leave open. Free PDF and talk video.

  • Shadowkube: enhancing Kubernetes security with behavioral monitoring and honeypot integration

    Converts compromised Kubernetes nodes into shadow honeypots in place, reporting 97.7% true-positive detection over 43 CVEs and 635 real attack attempts during public deployment. Open access, with live-deployment numbers rather than simulation.

  • HoneyFactory: Container-Based Comprehensive Cyber Deception Honeynet Architecture

    Container-based honeynet that deploys honeypots dynamically as an intrusion progresses, using a Gaussian hidden Markov model to estimate attack stage. Open access; evaluation is simulation-based.

  • LLM in the Shell: Generative Honeypots

    shelLM, an LLM-backed Linux shell honeypot reporting a 0.90 true negative rate against security experts asked to tell it from a real host. The paper that started the generative-honeypot line.

  • Container Orchestration Honeypot: Observing Attacks in the Wild

    High-interaction honeypot for exposed Docker and Kubernetes endpoints that drew attackers within minutes and produced 94 days of attack data, shared with the community. Free PDF from the authors' lab.

    open version
  • A Virtual Honeypot Framework

    Honeyd: simulating whole virtual hosts at the network level so one machine can present thousands. The foundational low-interaction honeypot paper, and still the mental model most tooling inherits.

    datedseminal
  • Know Your Enemy: Learning about Security Threats, 2nd Edition

    The Honeynet Project's collective account of running honeynets and reading what walked in. Badly dated. Out of print, though two sample chapters and the CD contents are still free.

    datedseminal
  • cowrie/cowrie

    The default SSH and Telnet honeypot. Emulated shell, a proxy mode fronting a real host, an experimental LLM backend. Python, actively developed.

    livingopen version
  • mushorg/conpot

    ICS/SCADA honeypot emulating industrial protocol stacks. Python, GPL-2.0. The usual starting point for OT deception.

    living
  • paralax/awesome-honeypots

    The long-running index of honeypot software by protocol and environment, with a section on anti-honeypot detection tools. Breadth over curation.

    living
  • telekom-security/tpotce

    Around thirty honeypot daemons bundled with Elasticsearch, Kibana and Suricata behind one installer. GPL-3.0.

    living
  • thinkst/opencanary

    Multi-protocol honeypot daemon light enough for a Raspberry Pi. BSD-3-Clause. The open counterpart to Thinkst's commercial Canary.

    living
  • The Honeynet Project

    501(c)(3) running since 1999. Institutional home of most of the open honeypot software listed here.

    livingopen version

Honeytokens 4

Planted credentials, files and records that alert when touched.

  • Active Defense and Deception Technology Against Insider Threats: A Serverless and Cryptographically Verifiable Framework

    Serverless cloud framework that plants generated honeyfiles, honeytokens and tracking beacons in PDF, DOCX and XLSX files to catch insiders, then revokes their cloud access automatically. In an emulated AWS setup it raised no alerts on 500 benign events from six simulated employees and issued revocations in about 2.2 seconds over 11 warm trials. The 100% detection claim rests on the authors' own scripted attacks with no trial count given, and Microsoft Word blocked the beacons. Peer reviewed in MDPI Informatics.

  • PHANTOM: polymorphic honeytoken adaptation with narrative-tailored organisational mimicry contextually convincing cyber deception at scale

    Generates honeytokens carrying organisation-specific naming and technology conventions, scoring 0.778 believability against 0.576 for templates. The authors concede the evaluation is 32 tokens scored by rules, not humans.

    contested
  • Canarytokens

    Hosted honeytoken generator, free: files, URLs, credentials, cloud keys that alert when touched.

    livingvendor-authored
  • We leaked LLM gateway api keys for 35 days. Here's what happened.

    A vendor planted 179 working LLM gateway keys in code examples across more than 40 GitHub organizations and logged their use at two decoy gateways, one mimicking LiteLLM. All 64 keys with a timed first request were tried within six hours, but only 18 keys sent prompts, and six of them drew 98.9% of 9,375 prompt requests. The figures come from one snapshot, the gateways ran 36 and 9 days rather than the headline's 35, and no data are released. Pairs with amazon-2026-artex-madpot.

    vendor-authored

LLM-backed honeypots 5

Decoys whose responses are generated by a language model rather than scripted.

  • Game of Phones: Harnessing Game Theory and LLMs in 'Bot Wars' to Counteract Phone Scams

    Combines game theory with LLM-driven conversational bots to engage phone scammers and waste their time, a form of conversational deception aimed at human adversaries. Whether it was tested on live scam calls or simulated scammers is unconfirmed, and that decides how far the result transfers. Springer LNCS chapter.

    paywalled
  • LLM Honeypot: Leveraging Large Language Models as Advanced Interactive Honeypot Systems

    Fine-tunes an open-weights model on captured attacker sessions to generate honeypot responses, then evaluates realism and deploys it live. Preprint; evaluation is thinner than shelLM's.

    preprint
  • SoK: Honeypots & LLMs, More Than the Sum of Their Parts?

    Systematizes both directions at once: LLMs used to build honeypots, and honeypots built for LLM attackers. Includes a taxonomy of honeypot detection vectors and a critique of how the area evaluates itself.

    preprint
  • 0x4D31/galah

    LLM-powered web honeypot: generates a plausible HTTP response to whatever arrives instead of emulating fixed applications. Go, Apache-2.0.

    living
  • beelzebub-labs/beelzebub

    Low-code deception runtime, Go, GPL-3.0. The interesting part is the MCP bait tooling: decoys aimed at AI agents rather than human intruders. A commercial product sits on top.

    livingvendor-authoredopen version

Deception against machine attackers 9

Deception designed for automated adversaries rather than human ones.

  • AgentTrap: Stateful Feedback Deception against Autonomous Penetration Testing Agents

    An LLM-driven honeypot endpoint that adapts its responses to an attacking agent and escalates to asking for the agent's own API key. Tested against eight open-source pentest agents with two models each, on a single SQL-injection target at three runs per configuration. Real-target success falls only from 95.8% to 79.2%, and keys leak in 18.8% of runs, all from CAI and PentestAgent. Other agents' models refused, or Strix's sandboxing blocked the leak. A four-page unreviewed preprint.

    contestedpreprint
  • Honeyquest for LLMs: Rethinking Cyber Deception for AI Attackers

    Twenty-one models against 174 reconnaissance queries, finding LLMs take deceptive bait far more often than humans, show no attention-diversion effect, and act on traps 73.4% of the time despite naming them in their reasoning.

    preprint
  • HoneyTrap: Deceiving Large Language Model Attackers to Honeypot Traps with Resilient Multi-Agent Defense

    Four cooperating agents divert jailbreak attempts into decoy responses, reporting a 68.77% average reduction in attack success while leaving legitimate queries intact. Preprint; the threat model is model-level, not network-level.

    preprint
  • Say Shibboleth: Fingerprinting the LLMs and Harnesses Behind Agentic Attacks

    Proposes honeypot services that hand an attacking agent credentials only after it names its model and writes free-text notes, plus probes that read the harness from how it masks secrets, enforces timeouts and schedules tool calls. Across about 800 lab runs it names the right model among 15 about 90% of the time and the right harness among four in 92% of runs. Those runs used containers on a private network, and agents told to give a fake name cut accuracy to 8%. Unreviewed v0.1 preprint extending pasquini-2024-mantis.

    preprint
  • Stinging the Swarm: Extracting Attacker Intelligence from a Twin-Controller OT Honeypot

    Describes Talos, an inline gate that screens each agent request to an industrial controller and reroutes agents it judges hostile to a mirror twin, where their attacks are mined for intelligence. Against 2,400 attacker agents in four concealment tiers of 600, no write reached the plant. The gate sent only 8% to 55% of each tier to the mirror and let 1,427 writes through, which a device-level check rejected, and every agent ran one abliterated Llama model against one simulated pump line. Unreviewed preprint with no repository link.

    preprint
  • Hacking Back the AI-Hacker: Prompt Injection as a Defense Against LLM-driven Cyberattacks

    Plants adversarial text in responses an attacking agent will read, disrupting it or compromising the attacker's own machine, with over 95% reported effectiveness. The cleanest statement of deception aimed at machines rather than people.

    preprintseminalopen version
  • The Deception Delta: Artifact for Adversarial Evaluation of LLM-Based Smart Contract Bytecode Forensics

    Artifact for a paper testing whether adversarially crafted smart-contract bytecode misleads LLMs used as forensic analysts: frozen model responses, judge outputs, bytecode and source inputs, gold labels and analysis scripts. It inverts this directory's usual direction, with the deceived machine on the defending side; read with prinos-2026-honeyquest-llms. CCS 2026 acceptance is claimed.

  • pasquini-dario/project_mantis

    The Mantis decoys as running code: tarpitted FTP, deliberately vulnerable web apps, weak telnet, injection payloads, reverse-shell listeners.

    living
  • Future of AI Honeypots

    Short essay predicting that agents will learn to spot honeypots and dodge fingerprinting, so traps should instead test for the mix of intelligence and speed that hiding would sacrifice. It reports no new measurements; the evidence is the author's 2024 honeypot, whose emulated Cowrie shell frontier models later saw through within seconds, plus figures cited from others. The author co-wrote pasquini-2026-shibboleth, whose fingerprinting approach this post expects to lose force. Individual blog post, not peer reviewed.

Counter-deception 1

Detecting, fingerprinting and evading deception from the attacker's side.

Empirical efficacy 4

Experiments and trials measuring whether deception imposes real cost.

  • Examining the Efficacy of Decoy-based and Psychological Cyber Deception

    The results paper from the Tularosa data: a four-condition design finding that decoys plus telling attackers deception is present changes behaviour most. The strongest peer-reviewed evidence that deception imposes real cost.

    seminal
  • The Tularosa Study: An Experimental Design and Implementation to Quantify the Effectiveness of Cyber Deception

    Over 130 red-team professionals in a two-day penetration task, with presence and disclosure of deception manipulated independently, plus questionnaires and physiological measures. The design paper that most later efficacy claims trace back to.

    seminalopen version
  • Cyber deception trials: what we've learned so far

    NCSC's December 2025 interim findings, from trials across 121 UK organisations and 14 commercial providers. Deception does surface hidden compromise. It is also not plug-and-play, and needs deliberate configuration to produce anything — which is the finding most vendor material omits.

  • Imposing a Cyber Penalty Against Attackers with Cyber Deception

    Readable summary of the Tularosa findings for practitioners: 52% of attacker commands targeted decoys, and exploit failures doubled under deception. Read this before the two formal papers.

Adversary cognition 5

Modelling attacker psychology, bias and decision making under deception.

  • A Cyber+Human Data Windfall to Revolutionize Cyberpsychology, Predictive Modeling, and Cyber Defense

    Overview of IARPA's ReSCIND programme, in which five teams spent 18 months testing whether attackers' cognitive biases can be measured and exploited. Over 2,300 participants from 59 countries took part, more than 450 of them cyber experts, with linked psychometric and host/network data released through IARPA. The claim that attack behaviour can be steered summarises the teams' own papers, where the effect sizes sit, and most participants were not professionals. Written by the programme's managers and paywalled; read with iarpa-rescind.

    paywalledopen version
  • Cyber expert feedback: Experiences, expectations, and opinions about cyber deception

    Questionnaire study of professional attackers finding they rarely expect deception, and that those who encounter it diverge sharply, some escalating effort and some abandoning it. Useful counterweight to averaged effect sizes.

    paywalled
  • Exploratory Analysis of Decision-Making Biases of Professional Red Teamers in a Cyber-Attack Dataset

    Re-analysis of red-team data for confirmation bias and framing effects, reporting that framing reduced attacker interaction with the network. Exploratory by the authors' own description, and behind a subscription.

    paywalled
  • Cyber Deception: Techniques, Strategies, and Human Aspects

    Ten chapters across game theory, AI and cognitive science — honeyfiles, ICS deception, breach detection. Two are open; the rest are behind Springer.

    paywalled
  • ReSCIND: Reimagining Security with Cyberpsychology-Informed Network Defenses

    US intelligence-community programme funding defenses that exploit attacker cognitive biases. Began 2023, three phases; solicitation closed.

    open version

Denial and deception theory 1

The pre-cyber literature on deception that the field keeps reimporting.

  • Cheating and Deception

    The deception typology cyber work keeps reimporting, often without citing it: masking, repackaging, dazzling, mimicking, inventing, decoying. Predates the field. Strong on structure, thin on evidence.

    datedpaywalledseminal

Frameworks and doctrine 7

Structured references for planning and running deception programmes.

  • Planning and Integrating Deception into Computer Security Defenses

    Three-phase model for planning deception that makes attacker bias an explicit design input and carries Bell and Whaley's six techniques into computer security. Widely borrowed from, rarely credited.

    seminalopen version
  • Using Cyber Decoys to Strengthen Detection and Response

    CISA's 22-page guide for critical infrastructure, aimed at intruders using valid credentials and living-off-the-land techniques. It follows MITRE Engage: start with tripwires on the EDR and identity tools already in place, and leave intelligence-gathering operations to teams that can handle the operational and legal risk.

    open version
  • NIST SP 800-160 Vol. 2 Rev. 1: Developing Cyber-Resilient Systems: A Systems Security Engineering Approach

    The cyber resiliency engineering framework. Cite this when a deception programme has to map to recognised controls.

    open version
  • MITRE Engage

    MITRE's adversary engagement framework: prepare, operate and understand phases over a goal/approach/activity matrix, with a starter kit. Successor to MITRE Shield, and the common vocabulary most deception programmes end up using.

    open version
  • 0x4D31/deception-as-detection

    Crosswalk from deception-based detections to ATT&CK techniques across Windows, Linux and macOS, with a worked deception story for each. MIT. Dormant for years, and nothing has replaced the mapping.

    datedliving
  • mitre/engage

    The Engage matrix as machine-readable data, plus the site tooling. Apache-2.0.

    living
  • HADES creates alternate reality to mislead hackers

    Moves an intruder into an altered environment to induce doubt rather than ejecting them. A press release, not a paper, and dated.

    dated

Moving target defense 6

Changing the attack surface over time; adjacent discipline to deception.

Deployment 1

Getting deception running and monitored in a real environment.

  • Is Cyber Deception a Must-Have This Time?

    Practitioner essay asking whether the AI era will finally make cyber deception a standard control, after earlier waves ended with standalone vendors being bought up. It brings no new data, but notes a 2026 industry briefing that ranked deception ninth of 11 recommended actions. His answer is mostly no, because attention, visible results and mandates have not changed, so he advises planting decoy credentials and MCP servers inside agent and identity projects already underway. Individual blog; compare ncsc-2025-deception-trials.

Datasets 2

Captured attack data released for reuse.

Foundations 3

The works everything else in the area cites.

  • Foundations of Cyber Deception: Modeling, Analysis, Design, Human Factors, and Their Convergence

    Twelve chapters across theory, human factors and application domains including ICS and UAV networks. Pitched at students entering the area.

    paywalled
  • Cyber Deception: Building the Scientific Foundation

    Thirteen chapters — adversary modelling, honeypotting, anonymity, forum analysis — assembled as a graduate seminar text. The first serious attempt to give the area a scientific spine.

    paywalled
  • Introduction to Cyberdeception

    The only general textbook on the subject. Impersonation, delays, fakes, camouflage, false excuses, on both offence and defense, with sustained attention to industrial control systems throughout rather than in one token chapter.

    paywalled

Trust and deception 1

Quantum-like treatments of trust, credibility and adversarial judgment.

  • A formulation of computational trust based on quantum decision theory

    Splits trust into objective and subjective components and uses interference terms to model how evaluations shift between isolated and comparative judgment. One of very few quantum-cognition papers aimed at a security-adjacent problem.

    paywalled

Peacetime international law 1

Sovereignty, non-intervention, countermeasures and state responsibility for deception below the threshold of armed conflict.

Active defense and computer-crime law 1

Hacking back, beacons and honeypots under computer-crime statutes such as the CFAA, and proposals to change them.

  • Avoiding the Pitfalls of Operating a Honeypot

    Lawyer's guide: phone-home tech on attackers' systems 'almost certainly' violates CFAA/state laws; honeypot domains risk trademark claims; GDPR/ECPA limit data collection; entrapment concerns may deter prosecution.

Privacy and wiretap law 3

Whether watching intruders in a honeypot is interception, and who is liable for what they do there.

  • Deploying Honeypots and Honeynets: Issues of Liability

    Examines civil and criminal liability of honeypot/honeynet administrators and the liability of attackers caught by them.

    paywalled
  • Liability and Ethics of Honeypots

    Downstream liability if a honeypot is used to attack others is unresolved and set by state law; operators may face copyright liability if attackers distribute contraband, so operators need to filter outbound traffic.

    dated
  • Use a Honeypot, Go to Prison?

    Reports DOJ attorney Richard Salgado's view that honeypot monitoring can be interception under the Wiretap Act; banners, the computer-trespasser exception and provider exception may help, but a honeypot built to be attacked may not qualify.

    dated

Data protection 3

GDPR and other data-protection rules applied to honeypots, decoys and botnet defense.

  • Deploying Honeypots and Honeynets: Issue of Privacy

    Analyzes what data EU-based honeypot operators may lawfully collect and retain, distinguishing low- vs high-interaction and research vs production honeypots.

    paywalled
  • Legal Issues of Honeynet's Generations

    Maps legal issues (privacy, liability, compliance) across honeynet data control, capture, collection and analysis functions under EU law.

    paywalled
  • Proactive Detection of Security Incidents: Honeypots

    EU agency study of honeypot tools with a short legal section flagging liability if a honeypot is used in attacks and advising CERTs to obtain jurisdiction-specific legal advice.

    dated

Ethics of defensive deception 1

What defensive deception owes attackers and bystanders where the law is silent: consent, harm and responsibility.

Surveys and taxonomies 4

The fastest way into a literature.

Courses 1

Structured teaching material.

  • Building Intrusion Detection Honeypots Training

    Twelve-plus hours on HTTP, SSH and RDP honeypots, honey tokens, and the monitoring around them. $497 with the companion book. Listed despite the price because free material at this depth does not appear to exist.

    livingpaywalled

Venues 3

Where this work is published and presented.

People 1

Researchers with substantial public bodies of work.

  • Curriculum Vitae for Neil C. Rowe

    Naval Postgraduate School; around 300 publications on deception in system defense, fake honeypots, and the ethics of deceptive operations. The CV links straight to full texts, which is why it is here rather than a departmental bio.

    livingopen version