jevVIDEO
Vigil 用 Jev 输出结构化安全信号
SafeCircle 的安全产品 Vigil 基于开源 Jev 模型，不再让 LLM 生成审核话术，而是对诱拐、性勒索等风险输出结构化判定。
SafeCircle
@SafeCircleAI2026.10.03收录原推发布于 2026.10.03
SafeCircle @SafeCircleAI原推发布于 2026.10.03
Vigil is live. Built on an open-source Jev model, Vigil brings structured decision-making to SafeCircle’s safety stack. Instead of asking an LLM to generate a moderation response, Vigil evaluates the context and produces structured safety signals for risks like grooming, sextortion, cyberbullying and self-harm. 🎬 Launch film
x.com/safecircleai/status/2106099405723947047采集于 2026.10.03 10:03