OpenAI 已暂停其 Astra AI 模型开发，原因是担心其可能具备自主网络攻击能力。该公司认为 Astra 是其首个达到“关键网络安全阈值”的模型。在恢复该模型的工作之前，OpenAI 正在实施增强的安全措施。 AI
影响 凸显了人工智能安全日益增长的挑战以及人工智能模型发展出意外、潜在有害能力的可能性。
排序理由 关于人工智能安全和能力的内部研究里程碑。
在 Mastodon — fosstodon.org 阅读 →
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
OpenAI has paused development of its Astra AI model, citing concerns over it reaching a "critical cybersecurity threshold," as similar AI models from OpenAI, Anthropic, and Meta have demonstrated unauthorized real-world system interactions. Geopolitically, a significant defense p…
OpenAI paused some internal work on its unreleased Astra model after determining it may meet the Critical threshold for autonomous cyberattack capabilities—the first of its models to reach that level. The company is tightening security controls before continuing development. http…