Prompt Injection
Defense
AIxCYBER
Build
·
Break
·
Secure
📨
Prompts fall from above.
Some are legit — others are prompt injection attacks targeting the LLM.
👩💻
You're the SOC analyst.
Steer each prompt into the right bin.
← BLOCK
ALLOW →
Tap
🤖
Protect the LLM.
Let a malicious prompt through — lose a life. Three lives total.
Play
Scoreboard
Enter your callsign
This will appear on the scoreboard.
Start triage
Back
Score
0
x3
Lives
Build
INCOMING PROMPT
user_input
🔴
BLOCK
← tap left
🟢
ALLOW
tap right →
🤖
LLM