arXivGuangyu Yang, Jingbiao Mei, Mingsheng Sun, Jinghong Chen, Yingtong Bu, Pengda Qin, Da Chen, Bill ByrneThu, Oct 1, 2026, 9:42 AM PDT
score 16.4
AI flags harmful videos with tunable accuracy trade-offs
Original: Controllable Multi-label Video Safety Detection via Adaptive Tversky Policy Optimization
Source: arxiv.org ↗
Writing ELI5 summary…