arXivBenedict Florance Arockiaraj, Alexander Feng, Jianxiong Cai, Xiaoyu ChengSun, May 17, 2026, 11:47 PM PDT
score 16.9
Researchers craft natural-sounding phrases that fool AI text models
Original: Universal Adversarial Triggers
Source: arxiv.org ↗
Writing ELI5 summary…