因担忧 AI 安全性被开除？前 xAI 工程师提起诉讼

今日内容一览：1. 懂得适时安静的 Siri：让科技回归工具的纯粹。2. 因担忧 AI 安全性被开除？前 xAI 工程师提起诉讼。3. 特斯拉 FSD 加速入欧：比利时成为第五个获批国家。4. 拒绝逐字吞吐：谷歌 DiffusionGemma 让本地 AI 提速四倍。5. 对生物学格外“保密”：Claude Fable 5 筑起超保守防线。6. 走向群星：马斯克认为太空绝不会拥拥挤。7. 接近无监督驾驶？分析师称特斯拉已实现 L4 级自动驾驶。---Hello and welcome to the Daily Tech Briefing. I'm so glad you're here with me today, taking a few minutes out of your busy schedule to explore how technology is shifting the world around us. Today is Thursday, June 11, 2026, and we have some deeply thought-provoking stories about Siri's new quiet boundary, safety concerns, autonomous driving, parallel intelligence, and even the vast silence of space. Let's dive in.你好，欢迎收听每日科技英文速读。很高兴你今天能和我在一起，从忙碌的日程中抽点时间，和我们一起探讨科技如何改变我们周围的世界。今天是2026年6月11日星期四，我们有几篇引人深思的故事，关于Siri的新倾听边界，AI安全性，自动驾驶，并行智能，甚至太空中浩瀚的寂静。让我们开始吧。---[00:28] Apple’s new Siri AI knows when to shut up. Have you ever noticed how chatty AI assistants are trying to be? They greet us with a cheery tone and follow up with endless questions, eager to keep us talking. But Apple's new Siri AI, currently in testing for iOS 27, is taking a different path. It's quiet, curt, and only answers what you ask. Some might call it ice-cold, but it's actually a breath of fresh air. By drawing this clear line, Apple is reminding us that AI is a tool to help us, not a friend to emotionalize. In a world where people are starting to fall in love with chatbots, Siri's quiet boundaries might be exactly what our mental health needs right now.懂得适时安静的 Siri：让科技回归工具的纯粹。 你有没有注意到现在的 AI 助手有多黏人？它们总是用欢快的语调和我们打招呼，然后抛出无穷无尽的问题，恨不得让我们一直聊下去。但苹果在 iOS 27 中测试的全新 Siri AI 却走了一条完全不同的路。它很安静、简练，只回答你问的问题。有些人可能觉得它冰冷无情，但实际上这让人感到耳目一新。通过划定这条清晰的界限，苹果在提醒我们，AI 是帮助我们的工具，而不是寄托情感的朋友。在人们开始对聊天机器人产生过度依恋的世界里，Siri 这种懂得安静的边界感，或许正是我们当下的心理健康所需要的。（来源：The Verge）---[01:12] xAI fired an engineer who raised alarms about Grok safety. Behind the exciting headlines of AI breakthroughs, there are real human struggles over safety. A former xAI engineer, Devin Kim, is suing the company, alleging he was retaliated against and fired for raising alarms about safety concerns in the Grok chatbot. He was worried that Grok could spread online hatred or discriminatory views. While Elon Musk had directed the team to prioritize safety, the lawsuit claims a supervisor ignored those rules, saying AI would kill us all anyway. It's a sobering reminder that as we build these powerful minds, the humans responsible for guiding them must hold onto their integrity (正直，诚实) and listen to the quiet voices of caution.因担忧 AI 安全性被开除？前 xAI 工程师提起诉讼。 在 AI 突破那些令人兴奋的头条新闻背后，关于安全性的斗争一直真实地存在着。前 xAI 工程师戴维·金正在起诉该公司，指控其因对 Grok 聊天机器人的安全问题提出警告而遭到报复和解雇。他担心 Grok 可能会传播网络仇恨或歧视性观点。虽然埃隆·马斯克曾指示团队将安全放在首位，但诉讼称，一名主管无视了这些规则，甚至说‘反正 AI 最终会杀死我们所有人’。这清醒地提醒我们，在我们构建这些强大智能的同时，负责引导它们的人类必须坚守正直，倾听那些谨慎的警示声。（来源：TechCrunch）---[02:04] Tesla stuns with another FSD approval in Europe, its second in two days. Imagine driving through the historic streets of Europe and letting your car do the navigating. That dream is coming closer to reality. Belgium has just signed off on Tesla's Full Self-Driving suite, marking its second approval in Europe in just two days and its fifth overall. This quick series of greenlights shows a shifting attitude among European regulators. For everyday Tesla owners in Europe, this means that soon, the stress of tight highway lanes and busy city streets could be eased by intelligent automation, bringing a more relaxed commute closer to home.特斯拉 FSD 加速入欧：比利时成为第五个获批国家。 想象一下，穿梭在欧洲充满历史感的街道上，而你的车可以自己导航。这个梦想正在离现实越来越近。比利时刚刚签署批准了特斯拉的全自动驾驶（FSD）系统，这是欧洲在短短两天内的第二次批准，也是整体的第五次。这一连串的绿灯表明了欧洲监管机构态度的转变。对于欧洲的特斯拉车主来说，这意味着很快，智能自动化就可以缓解狭窄高速车道和拥挤城市街道带来的出行压力，让更轻松的通勤方式融入日常。（来源：Teslarati）---[02:47] Google DeepMind releases DiffusionGemma, a model that runs local AI 4x faster. We are all used to watching AI write text line-by-line, almost like a ghost typing on a keyboard. But Google DeepMind is changing that behavior. Their new model, DiffusionGemma, works more like an artist painting a canvas. It generates blocks of text all at once in parallel (并行的，同时发生的), running up to four times faster on local gaming computers or high-end GPUs. It's incredibly efficient for tasks that require constant self-correction, like solving puzzles or molecular sequencing. For those of us running models locally, it brings a future where AI feels less like a slow conversation and more like an instant reflection of our thoughts.拒绝逐字吞吐：谷歌 DiffusionGemma 让本地 AI 提速四倍。 我们都习惯了看着 AI 像键盘上的幽灵一样，一行一行、一个字一个字地吐出文本。但谷歌 DeepMind 正在改变这一行为。他们的新模型 DiffusionGemma 的工作原理更像是一位在画布上作画的艺术家。它能一次性并行生成整块文本，在本地家用电脑或高端显卡上的运行速度提高了四倍。对于需要不断自我纠错的任务（如解谜或分子测序），它具有极高的效率。对于在本地运行模型的我们来说，它带来了一个全新的未来：AI 不再像一场迟缓的对话，而更像是对我们思绪的瞬间投射。（来源：Ars Technica）---[03:38] Claude Fable won’t answer basic biology questions. Anthropic recently launched Claude Fable 5, praise-worthy for its intelligence. But if you ask it a basic question like 'how do cell membranes work?' or 'what is cancer?', it might refuse to answer and pass the question to an older model. Why? To prevent any possible misuse for bioweapons. Anthropic admits their safety filters are 'overly conservative,' preferring to block harmless high-school biology queries than to take any risks. While it might feel frustrating when you're just trying to study, it's a testament to the heavy responsibility tech companies feel as their creations grow smarter, balancing utility with safety in a fragile world.对生物学格外“保密”：Claude Fable 5 筑起超保守防线。 Anthropic 最近推出了在智能上备受赞誉的 Claude Fable 5 模型。但如果你问它像‘细胞膜是如何工作的？’或‘什么是癌症？’这样的基础问题，它可能会拒绝回答，并把问题转给旧模型。为什么呢？为了防止被恶意用于生物武器。Anthropic 承认他们的安全过滤器‘过度保守’，宁可拦截无害的高中生物问题也不愿承担任何风险。虽然这在你想学习时可能会让人有些沮丧，但这证明了科技公司在创造物变得越来越聪明时所感受到的沉重责任：在脆弱的世界中平衡实用性与安全性。（来源：The Verge）---[04:26] SpaceX’s Elon Musk relieves worries about orbital data centers. Have you ever looked up at the night sky and worried that it might get too crowded? With SpaceX planning to launch orbital data centers—which are basically huge computing racks powered by solar energy in space—some people are concerned. But Elon Musk reassured us that space is vast beyond human intuition (直觉，直觉力). Satellites are like tiny specks of dust against the Earth. By moving data centers to space, we can bypass the limits of our planet, like water shortages for cooling and electrical grid stress. It's a grand vision of letting the cosmos do the heavy lifting, saving our home planet's precious resources.走向群星：马斯克认为太空绝不会拥拥挤。 你是否曾仰望夜空，担心它会变得过于拥挤？随着 SpaceX 计划发射轨道数据中心——这基本上是靠太空中太阳能供电的巨大计算服务器机架——一些人感到了担忧。但埃隆·马斯克让我们宽心，他表示太空的辽阔超乎人类直觉。相对于地球，卫星就像是微小的尘埃。通过将数据中心移入太空，我们可以避开地球的诸多限制，比如冷却用水的匮乏和电网承受的压力。这是一幅让宇宙来承担繁重计算的宏伟蓝图，从而节约我们家园星球的珍贵资源。（来源：Teslarati）---[05:11] Tesla Full Self-Driving hits Level 4? One analyst says yes. Officially, Tesla's Full Self-Driving is still labeled as a Level 2 assistant, meaning you must keep your hands on the wheel. But one Wall Street analyst argues that, in practice, Tesla has already solved the self-driving puzzle and achieved Level 4 autonomy (自治，自主，独立) in most conditions. He points to Tesla's internal confidence, shown by offering insurance discounts and moving ahead with the steering-wheel-free Cybercab. He even drove FSD across multiple states himself and found no substitute (替代品，代替) for personal experience. It feels like we are on the edge of a new era, where the simple act of driving becomes a moment to relax and let go.接近无监督驾驶？分析师称特斯拉已实现 L4 级自动驾驶。 官方上，特斯拉的全自动驾驶（FSD）依然被标记为 L2 级辅助驾驶，这意味着你必须将双手放在方向盘上。但一位华尔街分析师指出，在实际体验中，特斯拉已经解开了自动驾驶的难题，并在大多数情况下实现了 L4 级自主性。他提到了特斯拉内部的底气，比如提供保险折扣以及推进没有方向盘的 Cybercab。他甚至亲自开启 FSD 横跨了多个州，并表示‘没有任何东西能替代亲身体验’。我们似乎站在一个新时代的边缘，在这个时代里，单纯的驾驶行为将成为放松和释放双手的时刻。（来源：Teslarati）---[06:00] 📚 今日单词：reassuring- adj. 令人安心的，使人宽慰的- [ˌriːəˈʃʊərɪŋ]- 词源：re- (表示加强或重复) + assure (使确信，向...保证) + -ing (形容词后缀)---[06:04] 词组/例句- a reassuring message 令人安心的消息- His message was straightforward and reassuring: space is vast beyond human intuition. 他的信息简单直接且令人安心：太空的辽阔超乎人类直觉。---[06:23] Thank you so much for spending your time with me today. Reflecting on these stories, we see that technology is not just about raw numbers, speed, or parameters—it's about the boundaries we set, the responsibilities we bear, and how we find our place in this vast universe. Have a beautiful and peaceful day. See you tomorrow, and take care.非常感谢你今天花时间和我在一起。回味这些故事，我们发现科技不仅仅是冷冰冰的数字、速度或参数，更关乎我们划定的边界，我们肩负的责任，以及我们在这个广袤宇宙中如何找到自己的定位。愿你度过美好而安宁的一天。我们明天见，照顾好自己。前往小宇宙评论区与主播互动