Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
1
0
3
The smaller the font, the safer your VLM might be: typographic attacks reveal a strong link between text-image embedding alignment and vulnerability in vision-language models.