研究人员开发了 SafeGen-Bench，一个旨在评估图像条件文本到视频生成模型安全性的新基准。该基准解决了即使是来自安全文本和图像输入的有害内容生成挑战，定义了 10 个专注于时间序列和描绘行为的恶意类别。初步评估显示，当前模型在安全性方面存在困难，不安全得分高达 44.5%，并且单一模态的防护措施不足，在七个恶意类别中的失败率高达 80%。 AI
影响 凸显了当前文本到视频模型关键的安全漏洞，需要改进防护措施和评估方法以实现负责任的 AI 开发。
排序理由 该集群包含两篇详细介绍 AI 新研究和基准的学术论文，特别与视频生成相关。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
arXiv:2606.01481v1 Announce Type: new Abstract: With the rapid advancements in text-to-image diffusion models, generative video models (T2V models) like Sora can now produce short synthetic videos from a text prompt or an initial image. However, synthetic video generation -- espe…
arXiv:2606.02441v1 Announce Type: new Abstract: Identity-preserving video generation (IPVG) aims to synthesize high-fidelity videos that follow text prompts while faithfully preserving a reference identity. Despite recent progress, existing IPVG methods still struggle to balance …
Identity-preserving video generation (IPVG) aims to synthesize high-fidelity videos that follow text prompts while faithfully preserving a reference identity. Despite recent progress, existing IPVG methods still struggle to balance high-level semantic control and low-level identi…