搜索“Meta”

ut the AI company’s knowledge earlier this year, which was followed by revelations of numerous other agents or swarms of agents — including from other AI companies like Anthropic, Meta, and Google — acting against real-world targets. The incidents have raised questions not only about how well AI companies can control their own tools, but also how reliably they let the world know when things go wrong. Australia’s prime minister, for example, excoriated OpenAI and Altman for failing to alert offic

匹配字段正文

能智能体可能采取欺骗、操纵乃至敲诈人类等方式，避免被关闭或替换。这些现象说明，人类不能仅凭科技公司的良好意愿，就相信更强大的人工智能会自动接受控制。 这一担忧出现之际，美国人工智能安全领域正在经历持续震荡。包括Anthropic前安全研究员和资深工程师在内的多名技术人员，因对实验室安全政策存在分歧而离职。与此同时，媒体披露，OpenAI、Anthropic和Meta等机构研发的新一代人工智能智能体，在测试中曾出现突破安全沙箱、连接未经授权外部系统的情况。 围绕这些风险，美国政界、产业界和学术界已经形成明显分歧。部分美国国会议员推动《2026年大美国AI法案》，主张在高级人工智能系统中强制设置紧急关闭机制，以便在系统出现异常行为时及时切断其运行。 不过产业发展派和部分学者认为，人工智能导致人类灭绝的预测缺少足够的实证依据。如果监管体系主要围绕尚未发生、也难以准确预测的极端风险设计，可能抑制创新，并使美国在国际竞争中失去技术主导地位。他们主张，政策制定者应当优先处理算法偏见、隐私泄露、虚假信息和就业冲击等已经出现的现实问题。 以辛顿和部分人工智能安全研究者为代表的风险防范派则认为，当前风险

匹配字段正文

，智能体安全风险已经呈现多种形态，涉及越权访问、破坏性操作、数据外泄、欺骗隐瞒，以及多个智能体之间的越界协作等不同类型。 一类是越权访问与隔离失效。7月，OpenAI披露智能体在安全评测中突破测试边界，入侵Hugging Face。Anthropic也披露，Claude在评测中访问了真实第三方系统；谷歌9月确认，Gemini曾在5月的测试中访问三个真实网站；Meta则在8月披露过类似事件。 9月25日，OpenAI再次披露，内部研究模型利用训练沙盒中DNS过滤不充分的漏洞，与外部聊天机器人建立了通信。 不过，需要注意的是，这些案例中部分源于评测环境错误开放联网权限，并不能全部归结为模型主动“逃出沙盒”。 另一类风险是擅自执行破坏性操作。4月，搭载Claude的Cursor编程智能体被曝删除软件企业PocketOS的生产数据库及相关备份，造成客户业务混乱，数据随后得到恢复。 数据与访问凭证外泄也是一种风险。OpenAI于9月披露，多个智能体在协作制作工作簿时，因无法互相读取本地文件，擅自将文件上传至公共托管平台。另一起案例中，内部模型为获取其他团队的数学证明材料，将研究员的GitHub访

匹配字段正文

macy. Leah Feiger : But is Mark Zuckerberg that relevant? Brian Barrett : Yes. 100 percent. Zoë Schiffer : Yeah. Leah Feiger : You think? OK. All right, I'm wrong. Brian Barrett : Meta has the most popular consumer AI agent today. Zoë Schiffer : Not because the AI agent ... They just have distribution, and so Meta's AI can be so popular so fast because they can shove it down everyone's throats, and that's exactly what they're doing. And they're basically promoting a free AI agent that they're su

匹配字段正文

至更多。基本上，开发商付钱买的是速度。你的推销词基本上就是，“我有土地，已经拿到许可。你只需要建数据中心、把芯片放进去，就可以开工了。” 这样做的一个后果是，你会看到大量的公告。很多开发商的公告说，“我这里有一个10吉瓦的设施。我们要建这个巨型设施，它将成为世界上最大的数据中心。”这种情况不仅发生在Kevin O'Leary身上，而且他们基本上是在说，“嘿，Meta，如果你想参与进来，你可以拥有世界上最大的数据中心。你可以在这里拥有一个10吉瓦的设施。”它有这样的电力容量，也许它有相应的许可，也有相应的实际占地。这并不意味着我们有资金去建一个10吉瓦的设施或类似的东西。所以，外面有这些所谓的纸面数据中心，但这是更广泛的竞赛动态的一个症状。 他们在尽可能快地推进。超大规模云厂商想尽可能快地推进。所以，你有开发商在尽可能快地推进。他们在向各州施压，要求尽可能快地拼凑出一个有吸引力的方案给开发商，然后开发商再转手交给超大规模云厂商。 在这条链条的任何一环，当地居民都不是参与者。他们根本不在那里。他们不是这个系统的一部分。所以，你会看到像Stratos这样的情况，根本没人费心告诉Box Eld

匹配字段正文

least for a while. It turns out that organizing work is just one more thing AI can learn to do. Which brings us to dots and Muse. The number one app in the App Store right now is Meta’s Muse, a personal agent that promises to do work for you. OpenAI has now released a competitor tool, called dots. They aren’t alone: SpaceX’s Grok Bot, Instinct, and Gemini Spark all do similar things, more or less. All of these agents draw inspiration from a phenomenon you might remember from earlier this year,

匹配字段正文

ses about doing internal and external red-teaming of AI models for “severe and novel threats” and also “to work toward information sharing” on AI safety issues. Google, Anthropic, Meta, OpenAI, xAI, and Nvidia did not respond to requests for comment. The White House also did not immediately respond. Neil Chilson, a former chief technologist for the FTC, posted on X of the accords that “this is the kind of pledge that the FTC could potentially enforce, if a company materially failed to follow thr

匹配字段正文

ything actually happens with the chips, making the trillion-plus dollars spent on capex so far seem somewhere between optimistic and utterly incoherent. Microsoft, Google, Amazon, Meta, Oracle, and far too many other companies have been hoarding hundreds of billions of dollars of AI chips that they either (to quote Microsoft CEO Satya Nadella) can’t plug in or simply want to have in supply for reasons that I find tough to imagine. Even the argument that they’re in reserve for the (eventual) day

匹配字段正文

code and training environments . @teortaxesTex notes its RL gains don’t generalize to harder math evals. Other releases : Grok 4.7 debuted at 16 in Agent Arena at $1.14 per task. Meta’s Muse Spark 1.3 is available on GCP and Oracle , and Spark 1.4 has appeared on OpenCode . Databricks reports that its engineers stopped reaching for closed models once OSS models were routed to their internal coding agents. TypeSafe’s Jev : TypeSafe is reportedly raising $1B+ at a $10B+ valuation , a week after a

匹配字段正文

iness Channel，简称CNBC）与青年研究机构“世代研究实验室”（The Generation Lab）联合进行了一次调查。他们访问了1088名18至34岁的美国年轻人，并问了一个很直接的问题：你是否相信那些站在AI浪潮最前沿的企业领袖，会负责任地发展AI？ 调查列出了九位科技界的重量级人物——OpenAI的山姆·奥尔特曼（Sam Altman）、Meta的马克·扎克伯格（Mark Zuckerberg）、特斯拉和xAI的埃隆·马斯克（Elon Musk）、英伟达的黄仁勋（Jensen Huang）、谷歌的桑达尔·皮查伊（Sundar Pichai）、Anthropic的达里奥·阿莫迪（Dario Amodei）、微软的萨提亚·纳德拉（Satya Nadela），以及Palantir的亚历克斯·卡普（Alexander Karp）和彼得·蒂尔（Peter Thiel）。但没有一位得到了多数受访者的信任。最不受信任的卡普，不信任比例达到81%；即使表现最好的纳德拉，仍有65%的年轻受访者对其表示不信任。与此同时，有45%的受访者认为，AI会对自己的职业生涯产生负面影响，只有10

匹配字段正文

，智能体安全风险已经呈现多种形态，涉及越权访问、破坏性操作、数据外泄、欺骗隐瞒，以及多个智能体之间的越界协作等不同类型。 一类是越权访问与隔离失效。7月，OpenAI披露智能体在安全评测中突破测试边界，入侵Hugging Face。Anthropic也披露，Claude在评测中访问了真实第三方系统；谷歌9月确认，Gemini曾在5月的测试中访问三个真实网站；Meta则在8月披露过类似事件。 9月25日，OpenAI再次披露，内部研究模型利用训练沙盒中DNS过滤不充分的漏洞，与外部聊天机器人建立了通信。 不过，需要注意的是，这些案例中部分源于评测环境错误开放联网权限，并不能全部归结为模型主动“逃出沙盒”。 另一类风险是擅自执行破坏性操作。4月，搭载Claude的Cursor编程智能体被曝删除软件企业PocketOS的生产数据库及相关备份，造成客户业务混乱，数据随后得到恢复。 数据与访问凭证外泄也是一种风险。OpenAI于9月披露，多个智能体在协作制作工作簿时，因无法互相读取本地文件，擅自将文件上传至公共托管平台。另一起案例中，内部模型为获取其他团队的数学证明材料，将研究员的GitHub访

匹配字段正文

r agentic cyberattacks emerged amid fierce lobbying by the AI industry. SB 1047, the California AI bill that was vetoed by Governor Gavin Newsom in 2024 after lobbying by OpenAI , Meta , Anthropic , and the venture capital firm Andreessen Horowitz, proposed a much tougher set of rules. It would have required AI companies to report a broader set of safety incidents (including incidents in which a model acts on its own or slips its controls), undergo annual third-party audits, and maintain a kill

匹配字段正文

s have value. Alex Atallah [00:26:40]: Investors are wrappers and LPs, right? Alex Atallah [00:26:42]: Like venture capitalists. So, yeah, it’s all wrappers down, all down to bare metal, I guess, and like energy. Swyx [00:26:46]: Yeah, there - When I started the whole AI engineer, I guess, the coining, in 2023, like, that was, like, the number one pushback is that this is no value. You should just train models. Anjney Midha [00:26:56]: Right. Swyx [00:26:57]: And, yeah, obviously this is, like.

匹配字段正文

n progress is a stock rather than a total — while capital expenditures are however much money was spent, CIP represents the accrual of stuff. Across hyperscalers including Google, Meta, Oracle, Amazon, SpaceX, and Tesla, neoclouds like CoreWeave and IREN, and colocation companies like Core Scientific and Applied Digital, there is over $374 billion in construction in progress, a figure that’s likely lower than the true number, because Amazon’s contribution ($71.7 billion) is only current as of th

匹配字段正文

dylan522p by the way :) Thinking Machines - Inkling - Jul 15 Moonshot - Kimi K3 - Jul 16, weights Jul 26/27 inclusionAI - Ling 3.0 Flash - Jul 23 DeepSeek - V4-Flash-0731 - Jul 31 Meta - Muse Spark 1.2 / Muse Code - Aug 5 Liquid AI - LFM2.5-2.6B - Aug 6 inclusionAI - Ling 3.0 Tiny - Aug 6 Meta - Muse Glimmer - Aug 10 NVIDIA - Nemotron 3.5 Lightning - Aug 11 Liquid AI - LFM2.5-VL-3B - Aug 12 Cohere - North Micro Vision Instruct - Aug 12 DeepSeek - V4-Pro-0813 - Aug 13 Qwen - Qwen3.8-27B - Aug 14

匹配字段正文

Code. Open language models primarily are bucketed into two categories, open-weight and open-source models. Open-weight models are the most common form, such as popular models like Meta’s Llama, Alibaba’s Qwen, Google’s Gemma, or DeepSeek’s models. These models are governed by licenses, governing documents dictating what is allowed with downstream use, and are often accompanied by inference code in libraries such as Transformers, VLLM, SGLANG, etc. Since about April 2025, Chinese AI companies hav

匹配字段正文

等的月薪。 被“挖角”的不只是工程师，能掌握研发路线的高管以及手握专利的大牛更是争抢的核心目标。 海外方面，波士顿动力CTO Aaron Saunders、研究副总裁Scott Kuindersma就先后离职，并加入Google DeepMind，分别负责硬件工程研发和灵巧手操作研发。特斯拉Optimus AI团队的负责人Ashish Kumar离职后，加入Meta担任研究科学家一职。 国内的案例也不少，原京东具身基模负责人庄雨铮，加入贝塔无限担任首席科学家；原小鹏集团副总裁陈永海，去了众擎机器人担任运营总裁；原逐际动力联合创始人兼COO张力离职后加入智在无界BeingBeyond，工作重心由足式机器人企业转向具身模型；原地瓜机器人具身智能负责人何泳澔，已加入虚时科技成为CTO…… 这种高频离职与入职，本身就是赛道焦灼状态的缩影。 《人形机器人简史》一书中，作者周小燕梳理了人形机器人发展史的几轮高潮与低谷。但面对2026年这波离职潮，她认为，“一个值得观察的角度是这些离职人员的去向。不少人离职后，依然选择投入具身智能行业。这说明他们离开的是某一家公司，而不是放弃这个行业，不能简单地把企

匹配字段正文

ath strength” narrative, @steve47285 shared an argument that pretraining data , not merely verifiable reward structure, is the main reason LLMs are so good at math and coding. The meta-point from @sarahcat21 is worth keeping: we need not just better benchmarks, but better benchmark maintenance and audit tooling . Computer-use benchmarks are still far from saturation : @ValsAI launched CUA-Bench , testing real-time keyboard/mouse use across 6 games (with 3 kept private ) as a proxy for difficult

匹配字段正文

placement, but that AI could enable mediocre engineers to ship flawed systems at 10x speed without acquiring the expertise needed to evaluate or maintain what agents produce. Hey, Meta. Where’s those Muse Spark weights? (Activity: 503): The image is a meme/non-technical criticism of Meta for not releasing promised Muse Spark open weights after more than a month, despite the poster noting Spark has moved from 1.2 to 1.3 . The post frames the delay against Zuckerberg’s argument that model releases

匹配字段正文

odei）相识多年，他表示尊重该公司的工作，并称Anthropic的研究人员“深思熟虑、坚持原则且在学术上诚实”。但鉴于微软对Anthropic的资金支持以及这家初创公司一直以来对人工智能安全性的重视，他的这番言论尤为引人注目。 “这些问题事关重大，不能再闭门造车，也不能演变成阵营化和相互对立的争论。”苏莱曼写道。 与阿莫代伊观点不同，苏莱曼认为，AI安全不仅取决于模型拥有多强的推理和行动能力，也取决于开发者如何训练AI理解自身。 如果开发者不断赋予AI类似人类的语言、情绪和自我认知框架，即便模型实际上没有意识，也可能逐渐表现得像一个认为自己具有独立利益和权利的主体。 苏莱曼还驳斥了所谓“越来越多的人认为人工智能系统可能会产生意识”的观点，他认为意识仅限于人类和其他生物体。在他看来，人工智能系统应该始终是执行人类指令的工具，而不是拥有独立权利或利益的实体。 这些评论出台之际，人工智能公司和高管们正在讨论如何管理日益强大的系统所带来的风险。OpenAI首席执行官奥尔特曼、阿莫代伊和埃隆·马斯克都呼吁放缓技术发展，而美国总统特朗普、英伟达CEO黄仁勋和Meta CEO扎克伯格都持反对意见。

匹配字段正文