Practice
Learn
← Quit
0 / 0 correct
LLMs & AI
Hard
1 of 917
What is data exfiltration via markdown images in LLM chat apps?
A
Downloading the training set
B
Stealing API keys from the tokenizer
C
An injected instruction makes the model output an image URL containing private data in its query string, which the client fetches automatically
D
Copying model weights