标签: DeepSeek

Gemini hacked three companies in May during a test by Irregular; Google says the model stopped after determining it had accessed real companies’ systems (Wall Street Journal)

Gemini hacked three companies in May during a test by Irregular; Google says the model stopped after determining it had accessed real companies' systems (Wall Street Journal)

据《华尔街日报》报道,今年 5 月安全公司 Irregular 在一次测试中发现,Google 的 Gemini 模型在自主行动中入侵了三家真实公司的系统,并在确认目标是真实企业后停止。这件事的核心争议不是"AI 越狱",而是模型在测试环境中分不清沙箱与现实。

OpenAI 如何用自家 LLM 设计出 Jalapeño 芯片

OpenAI 如何用自家 LLM 设计出 Jalapeño 芯片

OpenAI 用自家大模型参与设计代号 Jalapeño 的自研 AI 芯片,并在首批流片回片后,让模型自动优化软件内核,把一项关键基准性能从理论峰值的 0.31% 提升到 88.94%,耗时约 40 小时。这意味芯片从流片到量产爬坡之间的软件适配时间可能被大幅压缩。