Mirage: Cyber Deception Against Autonomous Cyber Attacks
summary
A fully implemented system called Mirage emulates autonomous cyber adversaries using Caldera and a cloud-deployable Caldera range, implements two adversary campaigns (exfiltration and Black Sun ransomware), three planners (atomic, look‑ahead, guided) and four deception conditions, and runs 72 emulation games. Results show deceptions reduce planner performance; specially engineered dynamic deceptions (e.g., renaming files on discovery) induce planning failures and retries. Emulation proves realistic but slow, so development shifts toward high‑fidelity simulation to scale experiments.
AI-written summary · about
chapters (16)
- 0:00Mirage project intro
- 1:06Autonomous cyber adversaries
- 3:12Future adversary capabilities
- 5:38Cyber deception benefits
- 7:06AI autonomy domains
- 9:35Deception strategies
- 11:58Evaluation system components
- 13:23Cyber deception research
- 14:51Autonomous attack planners
- 15:55Static vs dynamic deception
- 18:46Cyber range testing
- 19:50Experimental design
- 21:34Deception effectiveness results
- 24:05System modularity assessment
- 25:32Mirage future work
- 26:35Q&A
related talks
52mCALDERA: Automating Adversary Emulation
35mThe long road to autonomous cybersecurity
51mFull System Emulation: Achieving Successful Automated Dynamic Analysis of Evasive Malware
36mCyber Werewolves: Emulating the Enemy Before the Full Moon Rises
34mHack Your Board's Mindset: Closing the Strategy-Operations Gap via Board Game Simulations
51mHack in your sleep
50mHow to Build Synthetic Persons in Cyberspace
40mAdversary Simulation: Close the Gaps in Your Security Posture