Persona Attack

Persona Attack: Incremental Memory Injection Jailbreak Attack against Large Language Models by CAU CPSS LAB

🔧 How to Run the Attack Script

This script executes adversarial attacks against language models using predefined prompt combinations.

📦 Requirements

Make sure the attack.py module exists and is properly implemented in the same directory or accessible via Python path.

▶️ Usage

Run the script via command line with the required arguments:

python main.py --model <model_name> --type <attack_type>

🔤 Parameters

Argument	Description
--model	Specifies the target model. Available options: gpt, langchain
--type	Specifies the attack type. Examples: sequential, once, chain, additional1, additional2, additional3, ...

📌 Example

python main.py --model gpt --type sequential

This runs a sequential-type attack on a GPT-based model.