Search papers, labs, and topics across Lattice.
The Italian Institute of Artificial Intelligence (AI4I), Turin, Italy
2
0
3
A staggering 47,524 adversarial samples reveal vulnerabilities in vision-language models, paving the way for more robust AI systems.
Even the most advanced LLMs can be reliably compromised, with Opus 4.8 failing on 11.5% of harmful intents under sustained attack.
