MultiTrust-X is a new 32-task, 28-dataset benchmark over 30 multimodal LLMs claiming that trustworthiness lags capability, that multimodality amplifies base-model risks, and that its RESA alignment method reaches state-of-the-art safety.
IEEE Transactions on Pattern Analysis and Machine Intelligence (2025)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Unveiling Trust in Multimodal Large Language Models: Evaluation, Analysis, and Mitigation
MultiTrust-X is a new 32-task, 28-dataset benchmark over 30 multimodal LLMs claiming that trustworthiness lags capability, that multimodality amplifies base-model risks, and that its RESA alignment method reaches state-of-the-art safety.