The KAP Trap Simulator
The lab's primary instrument is the KAP (Knowledge, Action, Prevention) Trap Simulator. It models common social engineering and crypto attacks to analyze the cognitive biases they exploit.
| Trap Simulation | Cognitive Triggers | Description |
|---|---|---|
| Fake Airdrop Simulation | Greed, Urgency (FOMO), Social Proof | Simulates an exclusive airdrop, prompting users to sign a malicious transaction to claim non-existent assets. |
| Malicious Token Approval | Trust Violation, Technical Obfuscation | Mimics a legitimate DeFi action, requesting a broad token approval that would grant a malicious contract control over user funds. |
| Impersonation Wallet Interface | Authority Bias, Pattern Interruption | Presents a UI nearly identical to a trusted wallet provider, designed to capture a seed phrase or password through a fake login prompt. |
Luredoor Case Files
The "Luredoor" project documents findings from these simulations:
- Case #001: Cracked Software: Demonstrated a user's willingness to bypass security for perceived value, quantifying the cognitive override of the "free" tag.
- Case #002: Jupiter NFT: Showed that urgency (FOMO) and social proof can systematically disrupt rational due diligence.
System Readout: AIXSELF Alignment
This adversarial research directly informs the AIXSELF safety architecture. By studying cognitive misalignment, we gather ground-truth data to build systems that are architecturally aligned. The goal is to design AI that is not just secure, but immune to the cognitive exploits that plague human systems.