Agentic backdoors via pretraining poisoning

2026-11-03来源 Hacker NewsEnglish

Article URL: https://arxiv.org/abs/2610.37741

Comments URL: https://news.ycombinator.com/item?id=50341334

Points: 49

# Comments: 13

原文：https://www.a8q9.cn/en/article/agentic-backdoors-via-pretraining-poisoning-9d40e7.html

相关文章

Apple Watch’s AI features for rewinding and recapping conversations begin beta t…2026-11-03
Yvonne Bajela is bringing $17M in fresh capital to Europe’s startup scene2026-11-03
I Drained a List From the Front. It Shifted Five Billion Elements.2026-11-03
Wiring iOS Vision Framework to a Core ML Embedding Model for Real-Time Semantic …2026-11-03
Content Security Policy Example: Start in Report-Only2026-11-03
Agentic AI för företag: vad det är, när det passar och hur du kommer igång2026-11-03