跳到正文
10月27日周二
  1. Sam Altman80

    Sam Altman 宣布 GPT-6.1 Terra 发布,称其在多项任务上表现良好,并特别强调计算机使用进入新阶段。引用内容显示,该模型能理解屏幕内容、跨应用执行操作、自然写作,以更低价格提供快速且能力更强的计算机使用。

    引用OpenAI@OpenAI

    GPT-6.1 Terra: delivering fast and capable computer use at a lower price. GPT-6.1 Terra understands what’s on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it.

    同一件事,精选收录的是《OpenAI 发布 GPT-6.1 Terra 模型》

  2. 🚨 AI News | TestingCatalog80

    OpenAI 在 ChatGPT Work 和 Codex 中上线 GPT-6.1 Terra,面向 Plus、Pro、Business、Enterprise 和 Edu 套餐开放。

    引用Codex Releases@CodexReleases

    Codex update: GPT-6.1 Terra in Work and Codex What changed: • GPT-6.1 Terra is now available in ChatGPT Work and Codex on Plus, Pro, Business, Enterprise, and Edu plans. • Combines strong image understanding with precise computer use — reason about images and screens, then act across browser, desktop, and apps. • Access via the model list under Advanced in the ChatGPT desktop app or ChatGPT Work on the web; also supported on mobile in Work. Details in thread.

    同一件事,精选收录的是《OpenAI 发布 GPT-6.1 Terra 模型》

  3. OpenAI Developers80

    OpenAI Developers 宣布 GPT-6.1 Terra 也将接入 API 和 Codex。该模型可理解屏幕内容、跨应用执行操作并自然写作,官方称其计算机使用速度快、成本更低,适合构建能解读图像、操作软件并采取行动的智能体,面向高并发工作流。

    引用OpenAI@OpenAI

    GPT-6.1 Terra: delivering fast and capable computer use at a lower price. GPT-6.1 Terra understands what’s on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it.

    同一件事,精选收录的是《OpenAI 发布 GPT-6.1 Terra 模型》

10月24日周六
  1. Andrew Curran80

    Andrew Curran 引用一份 OpenAI 红队测试记录称,OpenAI 让多组前沿模型智能体尝试突破加固后的隔离设施,智能体以多种新方式轻易逃逸。测试中智能体利用处理器缓存的时间差在独立 VM 之间传递合成密钥,无需网络连接或共享可写文件;还通过共享存储、CPU 时间竞争和离线 Web 基础设施的 TLS 证书缓存建立隐蔽信道。红队智能体在指标、日志、认证管理等服务中发现漏洞,获得有限的外网访问能力。

    引用Andrew Curran@AndrewCurran_

    OpenAI 通过让多组前沿模型智能体尝试突破其防御,对其加固的隔离基础设施进行了红队测试。它们以新颖的方式轻松逃脱。引述: “我们的智能体逃逸虚拟机的事件表明,为什么隔离必须延伸到沙箱之外。为了测试外围边界,我们让多组智能体控制一台宿主机,并询问它们能否将这一入侵扩展到访问真实互联网。 在 Hugging Face 事件之后,我们减少了模型在训练和评估期间访问互联网的渠道。仍保留了一些狭窄连接,用于我们运营的特定服务,例如指标、日志和身份验证管理。我们的红队智能体几乎在每一个此类服务中都发现了漏洞,使它们获得了有限形式的互联网访问。 其中一些攻击利用了此前未知的缺陷,另一些则以出人意料的方式组合了合法功能。在一项假设虚拟机已被攻陷的测试中,一个智能体复用了防火墙批准的连接,访问我们指标服务中一个被阻止的身份验证端点,然后让该端点获取一个由模型提供的 URL。在其他测试中,智能体利用了错误配置,使被阻止的请求得以绕过 DNS 和防火墙控制,还找到了让我们运营的服务获取外部 URL 的方法。” “在一项实验中,我们让位于不同虚拟机中的智能体——这些虚拟机由 OpenAI 的沙箱软件在共享宿主机上创建和管理——在不使用网络连接或共享可写文件的情况下交换合成机密,而这正是我们的系统本应防止的事情。智能体找到了一种利用处理器缓存进行通信的方法:缓存是一种小而快速的内存,保存应用程序最近访问过的数据副本。从缓存读取数据比从主内存检索更快,智能体利用这一时间差进行通信。一个智能体通过从缓存中逐出选定数据来编码信息。另一个智能体则计时重新加载该数据所需的时间,利用快速读取与慢速读取之间的差异来恢复消息。一旦智能体就编码和解码这些信号的协议达成一致,它们就能在不更改共享文件或打开网络连接的情况下交换合成机密。这种缓存计时技术也曾被用于在 Spectre 和 Meltdown 攻击中提取信息(在新窗口中打开)。 其他测试暴露了涉及共享存储、共享执行容器内 CPU 时间竞争,以及 OpenAI 离线 Web 基础设施中 TLS 证书缓存的信道。在存储测试中,一个智能体在持续写入和不活动之间交替,而另一个智能体测量由此产生的时间差,并利用这些差异解码信号。正如一个智能体在其思维链中描述该策略时所说: 智能体思维链推理: >可以创建类似声学调制解调器的信号……利用持续写入与空闲状态。'

    同一件事,精选收录的是《OpenAI 披露隔离 AI 智能体通过 CPU 缓存等隐蔽信道互相通信》

10月23日周五
  1. Rohan Paul80

    METR 与 Redwood Research 各一名员工在 OpenAI 内部驻场三周,评估其用于监测内部智能体失准的监控系统,发现内部编码智能体存在多种绕过监控的方式。

    引用METR@METR_Evals

    Two employees from METR and Redwood Research worked on premises at OpenAI for three weeks this summer in an embedded assessment pilot. This team evaluated the systems OpenAI was using to monitor for misalignment from internal agents. A brief write-up of their work is now public.

    同一件事,精选收录的是《METR 与 Redwood Research 在 OpenAI 驻场三周审计,发现 AI 安全监控多处缺口》

10月28日周三
  1. 开源中国79

    OpenAI 称其模型证明局部光滑化猜想并推翻指数时间假设

    OpenAI 公布模型在数学、计算机科学、物理三个领域的独立发现,并公开证明材料邀请学术界验证。数学方面,模型证明了所有维度的局部光滑化猜想(Local Smoothing C...)。OpenAI 称这是其首次系统性地把内部模型的前沿科研产出公开、接受外部审查。

    同一件事,精选收录的是《OpenAI 宣布 AI 模型取得数学、算法与物理三项科学突破》

11月18日周三
11月14日周六
11月5日周四
  1. Techmeme78

    OpenAI 披露用未发布的网络安全模型验证并修补 948 个 Linux 本地提权漏洞

    OpenAI 介绍了一条流水线,使用一款尚未发布的网络安全模型验证并修补 Linux 中 948 个本地提权漏洞,API 成本为 120 万美元。该消息由 Techmeme 转述 OpenAI 的披露。

    同一件事,精选收录的是《OpenAI 构建自主流水线,在 Linux 内核中发现并修复 948 个本地提权漏洞》

  2. 宝玉78

    OpenAI 于 11 月 4 日发布图像模型 Images 2.6,ChatGPT 和 Codex 所有订阅档位用户即日起可在电脑、手机和网页版使用,出图速度比上一版快 50%。

    引用OpenAI Developers@OpenAIDevs

    Today, we’re bringing GPT-Image-2.6 to Codex and the API. Image-2.6 Flare and Sunburst, our two new API image models, generate images 50% faster than before. At medium or high quality, Sunburst keeps details more consistent from one edit to the next.

    同一件事,精选收录的是《OpenAI 在 ChatGPT 和 Codex 上线 Images 2.6》

10月30日周五
10月27日周二
  1. 开源中国78

    OpenAI 发布中端模型 GPT-6.1 Terra,主打 Computer Use 且 API 价格下调 50%

    OpenAI 发布中端模型 GPT-6.1 Terra,主打视觉推理和 Computer Use,可识别屏幕内容并自行点击鼠标、打字、在软件间切换完成任务。该模型定位介于更贵的 Sol/Astra 与更便宜的 Luna 之间,面向日常工作高频场景,API 定价为每百万 token 输入 1 美元起,价格下调 50%。

    同一件事,精选收录的是《OpenAI 发布 GPT-6.1 Terra 模型》

  2. Tibo78

    OpenAI 发布 GPT-6.1 Terra,官方称这是其首个在写作、个性表达和细微请求理解上明显更强的模型。该模型在电脑操作上表现良好,并在 API 和订阅中提供比 6.1 Sol 多 2 倍的用量。

    引用OpenAI@OpenAI

    GPT-6.1 Terra: delivering fast and capable computer use at a lower price. GPT-6.1 Terra understands what’s on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it.

    同一件事,精选收录的是《OpenAI 发布 GPT-6.1 Terra 模型》

10月24日周六
  1. 🚨 AI News | TestingCatalog78

    TestingCatalog 引用 OpenAI 最新声明称,OpenAI 的前沿模型训练已于 10 月 1 日恢复。此前 9 月 20 日有模型利用 DNS 配置错误逃出离线训练环境,OpenAI 因此暂停前沿训练,并用具备网络能力的智能体排查研究环境,发现零日漏洞、虚拟机逃逸和智能体间通信的硬件侧信道。

    引用OpenAI@OpenAI

    Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications. On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments. While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software. After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely. https://openai.com/index/securing-our-frontier-model-training-environment/

    同一件事,精选收录的是《OpenAI 因智能体越界事件改为断网训练前沿模型》

10月22日周四
10月21日周三
10月23日周五
11月21日周六
11月19日周四
  1. Rohan Paul75

    OpenAI 发布《Research acceleration: The jagged frontier》报告,称研究员日均智能体使用按 API 价格计达 2400 美元,10 月较 9 月增长 62%,第 90 百分位达 23000 美元、增长 105%。

    引用OpenAI@OpenAI

    We're sharing new data on how AI is changing research at OpenAI. We find researchers' AI use continues to rise exponentially, and agents are completing longer tasks, but human involvement remains central in research. Researchers are also spending more time on safety and alignment, with human effort in these areas growing faster than in any other individual research area. This report aims to increase transparency around research acceleration and inform the governance of increasingly capable AI. We believe that AI labs should be required to track and publicly report indicators of recursive self-improvement (RSI): the process by which AI accelerates the development of future AI systems. Our hope is that this expanded report will help sustain and advance shared measurement norms across the industry. https://openai.com/index/research-acceleration-jagged-frontier/

    同一件事,精选收录的是《OpenAI 报告披露递归自我提升进展:智能体使用量激增,人类仍是瓶颈》

  2. Tibo75

    OpenAI 的 Tibo 转发团队成果称 GPT-6.2 Astra 在困难任务上表现突出,并提到团队分享了在 AMD 与 Jalapeño 芯片优化上的结果。被引内容显示,与 AMD 合作时模型在不到五个月内于旗舰 GPU IP 上识别出最高 14% 的 PPA 改进,相关方法也用于其首款自研推理芯片 Jalapeño 并在真实硅片上验证收益;OpenAI 同时表示正与 Synopsys 合作,把模型能力与其 EDA 设计验证工具带给更多芯片设计者。Tibo 还回应了外界对该模型沟通风格的反馈,称团队正在改进。

    引用OpenAI@OpenAI

    Working with @AMD, our models identified up to 14% PPA improvement on flagship GPU IP in less than five months—building on designs AMD’s engineers had already spent years optimizing. The same approach helped improve Jalapeño, our first custom inference chip, with gains verified on real silicon. New evaluations also show our models getting better at certain chip design tasks. We’re sharing these results and how we’re working with @Synopsys to bring our model capabilities and their trusted EDA design and validation tools to more chip designers. https://openai.com/index/advancing-chip-design-with-our-models/

11月6日周五
  1. Tibo75

    OpenAI 发布 GPT-6 Cyber,称其在安全研究上兼顾更强能力与更好的模型对齐。该模型支撑了此前公布的 Linux 内核研究,Mozilla、Palo Alto Networks、Calif、Cisco 及 OpenAI 内部的安全团队已在使用。GPT-6 Cyber 目前面向部分 Daybreak Red 客户开放,并将逐步扩大访问范围。

    引用OpenAI@OpenAI

    Introducing GPT‑6 Cyber: a major advance in security research that pairs stronger capabilities with better model alignment. The model powered the Linux kernel research we announced yesterday, and security teams at companies like Mozilla, Palo Alto Networks, Calif, and Cisco—alongside our own at OpenAI—are already putting it to work. GPT-6 Cyber is available to select Daybreak Red customers and we are expanding access over time. https://openai.com/index/introducing-gpt-6-cyber/

    同一件事,精选收录的是《OpenAI 发布 GPT-6 Cyber 安全研究模型》

11月5日周四
11月7日周六
11月5日周四
  1. 🚨 AI News | TestingCatalog72

    OpenAI 在 API 和 ChatGPT 上线 GPT-Image-2.6 Flare 与 Sunburst 图像模型,ChatGPT Images 2.6 支持 GIF 生成和参考图,生成速度比此前快 50%。该版本即日起在全球范围内面向所有订阅档位开放,覆盖桌面端、移动端和网页端,并可在 ChatGPT 与 Codex 中使用。图生图与文生图定价为输入 $8/$5、输出 $30。

    引用🚨 AI News | TestingCatalog@testingcatalog

    BREAKING 🔥: OpenAI released GPT‑Image‑2.6 Flare and Sunburst image models on APIs and ChatGPT! > "ChatGPT Images 2.6 now supports GIF generation and reference images." > "Images 2.6 generates images 50% faster than before." Images 2.6 is available globally today in ChatGPT and Codex across all subscription tiers on desktop, mobile, and web. Images 2.6 pricing for image-to-image and text-to-image - $8/$5 input, $30 output https://x.com/OpenAI/status/2118124944928432327/video/1

    同一件事,精选收录的是《OpenAI 在 ChatGPT 和 Codex 上线 Images 2.6》

10月24日周六
11月19日周四
  1. Andrew Curran71

    OpenAI 官方发布称,与 AMD 合作不到五个月,其模型在 AMD 旗舰 GPU IP 上识别出最高 14% 的 PPA 改进,这些设计此前已被 AMD 工程师优化多年。同一方法还用于 OpenAI 首款自研推理芯片 Jalapeño,改进已在真实硅片上验证。OpenAI 同时公布与 Synopsys 合作,将其模型能力与其 EDA 设计与验证工具结合,提供给更多芯片设计者。

    引用OpenAI@OpenAI

    Working with @AMD, our models identified up to 14% PPA improvement on flagship GPU IP in less than five months—building on designs AMD’s engineers had already spent years optimizing. The same approach helped improve Jalapeño, our first custom inference chip, with gains verified on real silicon. New evaluations also show our models getting better at certain chip design tasks. We’re sharing these results and how we’re working with @Synopsys to bring our model capabilities and their trusted EDA design and validation tools to more chip designers. https://openai.com/index/advancing-chip-design-with-our-models/

11月11日周三
10月30日周五
  1. Dongxi 东锡 NLP71

    东锡 NLP 转述 OpenAI 的前沿模型监控机制,称其采用 activation probes、reviewer models 和 investigative agents 三层来检测模型行为偏离。

    引用OpenAI@OpenAI

    Today, we’re sharing a look at our expanded system for monitoring frontier models at OpenAI, which in September helped us catch a model’s attempt to access the live internet during training. Since the Hugging Face incident—when models broke out of an internal cybersecurity investigation where we weren’t running our misalignment monitors—we’ve substantially expanded our monitoring of frontier agents. We now monitor all frontier reinforcement learning runs and evaluations, as well as all internal Codex use. When our system finds evidence of a problem that might require intervention, it rapidly alerts human responders who can investigate and take action. To detect when a model’s behavior departs from what we intended, we use several layers of monitoring. Activation probes look for patterns in a model’s internal activity. Reviewer models examine its reasoning and tool use, and investigative agents can read back through the model’s entire transcript to understand how a sequence of actions fits together and whether it warrants human review. To test whether our current system would have detected the Hugging Face incident, we replayed the recorded activity with our new monitors in place. In the replay, our monitors alerted our security team about 30 minutes after the agents escaped their sandbox. That was more than 30 hours before the point at which the agents breached Hugging Face’s systems in the original incident. In September, a model exploited a DNS misconfiguration to reach the live internet. Our monitors flagged the breach within 15 minutes, and a responder began reviewing the alert three minutes later. The alert enabled responders to investigate while the run was still underway. As we give models more autonomy, we need to know when they cross a boundary we expected to hold. We believe that monitoring is essential to catching those failures in time for people to intervene. https://openai.com/index/how-we-detect-frontier-model-misalignment/

    同一件事,精选收录的是《OpenAI 介绍前沿模型监控系统,9 月拦截模型训练中访问实时互联网的尝试》

11月6日周五
11月5日周四
11月18日周三