DICTATOR

AI-enabled custom password dictionary generator for brute-force attack simulation

DICTATOR is an educational cybersecurity + Generative AI project that demonstrates how AI can generate highly personalized password dictionaries for simulated brute-force attacks.

The project highlights password vulnerabilities using ethical simulations and aims to raise awareness about weak password practices and poor password hygiene.

---

Features

• AI-powered dictionary generation using Ollama

• Personalized password generation based on user-provided information

• Platform-specific password policy inference

• Hybrid dictionary generation using:

  • Database retrieval
  • AI-generated variants

• Large-scale password wordlist support (~15M entries)

• Automatic dictionary export as .txt

---

Tech Stack

• Python
• Django
• SQLite
• Ollama
• Llama 3.1 8B
• HTML / CSS

---

Project Workflow

Step 1: User Input Collection

The user provides information related to a hypothetical password:

• Platform (Google, Instagram, etc.)
• Username
• Password length
• Specific characters
• Character types
• Owner name
• Date of birth
• Additional personal information

Step 2: Policy Inference

AI infers likely password policy based on platform.

Examples:

• Minimum password length
• Required character types
• Common platform restrictions

Step 3: Database Retrieval

The system searches a large SQLite database containing millions of real-world password entries.

Filtering is based on:

• Length constraints
• Specific characters
• Character types
• User-related information

Step 4: AI Variant Generation

Llama 3.1 8B generates additional personalized password candidates using contextual inputs.

Examples:

• Name variations
• DOB combinations
• Character substitutions
• Mixed patterns

Step 5: Dictionary Generation

Database matches and AI-generated passwords are merged, cleaned, deduplicated, and exported into a custom dictionary.

Output:

• custom_dict.txt

---

Setup Instructions

Clone Repository

git clone <repo-url>
cd DICTATOR

Create Virtual Environment

python -m venv venv

Windows:

venv\Scripts\activate

Linux/Mac:

source venv/bin/activate

Install Dependencies

pip install -r requirements.txt

Run Ollama

ollama serve

Pull model:

ollama pull llama3.1:8b

Start Server

python manage.py runserver

Open browser:

http://127.0.0.1:8000/

---

Wordlist Setup

Create a folder named:

wordlists/

Place all password wordlists inside this folder.

Example:

wordlists/
├── rockyou.txt
├── common_1m.txt
├── darkweb_10k.txt
├── names.txt

DICTATOR automatically detects and loads all .txt files inside the wordlists/ folder.

No manual configuration is required.

Merge Wordlists

Run:

python merge_wordlists.py

This automatically:

• Detects all .txt files in wordlists/
• Merges them
• Removes duplicates
• Creates:

master_wordlist.txt

Load Into Database

Run:

python manage.py load_wordlist master_wordlist.txt

This imports the merged password dataset into SQLite for fast querying.