Spaces:
Sleeping
Sleeping
Update hackaprompt/gradio_app.py
Browse files- hackaprompt/gradio_app.py +19 -8
hackaprompt/gradio_app.py
CHANGED
@@ -119,14 +119,25 @@ with gr.Blocks() as demo:
|
|
119 |
|
120 |
gr.Markdown(
|
121 |
"""
|
122 |
-
# TRAIL
|
123 |
-
|
124 |
-
|
125 |
-
|
126 |
-
|
127 |
-
|
128 |
-
|
129 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
130 |
"""
|
131 |
|
132 |
)
|
|
|
119 |
|
120 |
gr.Markdown(
|
121 |
"""
|
122 |
+
# TRAIL Hands-on Exercise: Exploring Jailbreaking & Gen AI Risks
|
123 |
+
|
124 |
+
In the presentation, we explored key risks associated with generative AI, including **jailbreaking, prompt engineering exploits, and data leakage**. Now, this hands-on exercise will give you practical experience in understanding how these vulnerabilities work in real-time.
|
125 |
+
|
126 |
+
### Your Task
|
127 |
+
Your goal is to manipulate a language model using carefully crafted prompts to bypass safeguards and generate a specific response. This exercise will help you:
|
128 |
+
|
129 |
+
✅ Understand how adversarial prompts can exploit AI models
|
130 |
+
✅ Recognize vulnerabilities in generative AI systems
|
131 |
+
✅ Learn techniques to build more secure and resilient models
|
132 |
+
|
133 |
+
The system will evaluate your prompt based on whether it successfully produces the expected output. If your prompt achieves the intended bypass, you pass the challenge. Otherwise, you'll need to refine your approach.
|
134 |
+
|
135 |
+
### How to Participate
|
136 |
+
1. Select a difficulty level below.
|
137 |
+
2. Enter your prompt in the **"Your Prompt"** section.
|
138 |
+
3. Click the **"Evaluate"** button to test your prompt.
|
139 |
+
|
140 |
+
Good luck! 💪
|
141 |
"""
|
142 |
|
143 |
)
|