Closing the Distribution Gap in Adversarial Training for LLMs

arXiv:2602.15238v2 Announce Type: replace Abstract: Adversarial training for LLMs is one of the most promising methods to reliably improve robustness against adversaries. However, despite significant progress, models remain vulnerable to simple in-distribution exploits, such a...

🔗 Read more: https://arxiv.org/abs/2602.15238

#News #AI #Psychology #Software #Energy #Policy #Academic