AIGC宇宙 AIGC宇宙

文章列表

​OpenAI宣布2025年开发者大会将在旧金山举办,预计吸引1500多名开发者参与

OpenAI 已正式公布了其下一届开发者大会(DevDay)的时间和地点,定于2025年10月6日在旧金山举行。 这次大会将吸引超过1500名开发者参加,预计将成为迄今为止规模最大的一次开发者活动。 此次 DevDay 的议程丰富多彩,包含多个重要环节。
6/27/2025 3:00:23 PM
AI在线

扎克伯格亲自出马!Meta重金挖角OpenAI三大AI顶尖专家

Meta成功从OpenAI挖走三位顶级人工智能研究员:卢卡斯·贝耶尔(Lucas Beyer)、亚历山大·科列斯尼科夫(Alexander Kolesnikov)和翟晓华,引发业界广泛关注。 这三位研究员在机器学习和计算机视觉领域享有盛誉,以开发可扩展图像模型和视觉转换器(ViT)架构的突破性工作而闻名。 值得注意的是,他们此前曾在Google DeepMind工作,并于2024年底刚刚参与在苏黎世开设OpenAI办事处。
6/27/2025 10:00:52 AM
AI在线

​Gemini 将取代 Google 助手,Android 用户迎来新体验

近日,Google 宣布即将推出的 Gemini 功能将会在 Android 设备上取代 Google 助手。 根据 Android Police 获得的一封内部电子邮件,Gemini 的更新将在7月7日开始推送。 这项更新将允许用户在关闭 Gemini 应用时,仍然可以通过该 AI 助手来控制手机的电话、信息、WhatsApp 等应用。
6/27/2025 10:00:51 AM
AI在线

Gemini 即将取代 Google 助手,隐私保护新模式来袭!

Google 近期宣布,将在7月7日推出一项重大更新,让其全新的 AI 助手 Gemini 成为用户手机的智能助理。 这一更新允许 Gemini 在用户的手机上执行多项任务,比如拨打电话、发送信息、使用 WhatsApp 等,且无需开启 Gemini 应用。 听起来是不是很方便?不过,随着这一变化,许多用户对隐私问题表示担忧。
6/27/2025 10:00:51 AM
AI在线

杀疯了,这个 AI+Python 智能体救了看门老大爷!

最近AiPy很火,我用了有一段时间了,感觉严重影响睡眠,感觉这东西容易上瘾,今天推荐给大家。 AiPy它是基于AI Python,官方的说法是他们给大模型装上了手脚,也就是说,用大白话就能自动操作电脑、控制手机、甚至连家里的智能灯泡也能控制,听起来有点逆天。 然后AiPy是个开源项目,有命令行版和客户端。
6/27/2025 9:29:45 AM
老朱2000

五款大模型考「山东卷」,Gemini、豆包分别获文理第一名

果然,高考已经快被 AI 攻克了。 近日,5 款大模型参加了今年山东高考,按照传统的文理分科方式统计:豆包 Seed 1.6-Thinking 模型以 683 分的成绩拿下文科第一,Gemini 2.5 Pro 则凭借 655 分拔得理科头筹。 测评来自字节跳动 Seed 团队。
6/27/2025 9:24:04 AM
机器之心

​Meta 挖角 OpenAI 顶级研究员,AI 超级智能团队再添猛将

在 AI 领域的竞争日益激烈之际,Meta 公司于今天凌晨宣布,成功挖角 OpenAI 的顶级研究员 Trapit Bansal,加入其新成立的 AI 超级智能部门。 根据 Techcrunch 的消息,OpenAI 的发言人 Kayla Wood 证实了 Bansal 的离职,Bansal 的 LinkedIn 页面也显示他已不再是 OpenAI 的员工。 图源备注:图片由AI生成,图片授权服务商MidjourneyTrapit Bansal 自2022年起在 OpenAI 工作,他被认为是该公司强化学习的重要推动者之一,同时也是 OpenAI 首个 AI 推理模型 o1的基础贡献者。
6/27/2025 9:00:41 AM
AI在线

告别复杂SQL!用Spring AI + DeepSeek构建自然语言查询系统

项目概览现代应用越来越多地采用自然语言交互界面,让用户更轻松地操作系统。 这在数据查询场景中尤为实用,非技术人员可以直接用日常语言提,文本转 SQL 聊天机器人正是这样的典型应用。 它在人类语言和数据库之间搭建了沟通桥梁。
6/27/2025 2:00:00 AM
PIG AI

Hengbot 推出 Sirius 机器狗:集成 OpenAI,可跳舞、踢足球

Hengbot 推出 Sirius 机器狗,集成 OpenAI 大语言模型,可跳舞、踢足球,还能语音对话。预售价 1299 美元,预计秋季上市。#机器人##AI#
6/26/2025 11:34:48 PM
潞源(实习)

AI 硬件新战场!出门问问创始人揭秘未来科技与市场挑战

在北京的一场发布会上,出门问问的创始人兼 CEO 李志飞宣布推出全新 AI 硬件产品 ——TicNote,这是一款具备智能体 AI 系统的设备。 TicNote 的厚度仅为3mm,可以通过磁吸方式直接贴在手机背后,搭载了基于 DeepSeek-R1等大语言模型的 Shadow AI 技术,支持 AI 转写、总结、思维导图生成等功能,录音时间更是超20小时,覆盖120多种语言。 李志飞在发布会上表示,通用人工智能(AGI)的目标是解放80% 的人工工作,而在软件领域,这一目标已经实现。
6/26/2025 6:01:29 PM
AI在线

High-Level Call Between OpenAI and Microsoft! The Future of Their Cooperation Remains Uncertain

Amid the intensifying competition in the field of artificial intelligence, Sam Altman, CEO of OpenAI, recently had a phone call with Satya Nadella, CEO of Microsoft, discussing future cooperation. This information was revealed in Altman's podcast interview on Tuesday, where he mentioned that the conversation with Nadella mainly focused on how to revise their investment terms and equity issues for the future.It is known that Microsoft is a major investor in OpenAI, and recent differences have arisen between the two parties regarding investment details, particularly about the scale of Microsoft's future shareholding. If no consensus can be reached on these key issues, Microsoft even considered suspending further discussions with OpenAI.
6/26/2025 5:01:43 PM
AI在线

Google Launches Imagen4: Breaking the Text-to-Image Generation Bottleneck, Gemini API Empowers Text-to-Image

Recently, Google officially launched its latest text-to-image model **Imagen4** through the Gemini API, marking a significant milestone in the field of generative AI (AIGC). According to Google's official blog and community feedback, Imagen4 has made breakthroughs in generating text within images, solving long-standing technical bottlenecks in AIGC and providing developers with high-quality tools for visual content creation.It is reported that the model comes in two versions: **Imagen4** and **Imagen4Ultra**, priced at $0.04 and $0.06 per image respectively. Currently, paid previews are available on the Gemini API and Google AI Studio, with some free trial spots open.Compared to its predecessor Imagen3, Imagen4 significantly improves text rendering quality, supports image generation up to 2K resolution, and covers a diverse range of artistic styles from realistic to abstract.
6/26/2025 5:01:43 PM
AI在线

OpenAI Fully Transitions to Rust for Refactoring Codex CLI to Improve Performance

Recently, OpenAI announced that it will restructure its Codex CLI tool, completely abandoning TypeScript and switching to the Rust language. This decision aims to provide developers with a more efficient and stable AI terminal interaction experience. Codex CLI was initially designed to simplify developers' interactions with AI in the terminal, using TypeScript and the React-based Ink framework.
6/26/2025 5:01:39 PM
AI在线

ChatGPT iOS App Monthly Downloads Exceed 30 Million, Surpassing All Social Apps

The iOS app of ChatGPT has been downloaded 29.6 million times in the past 28 days, becoming the most popular app worldwide. This achievement has made ChatGPT surpass the combined downloads of four major social apps—TikTok, Facebook, Instagram, and X—which had approximately 32.9 million downloads during the same period, with a gap of 10.6%. Although social apps have been on the market for a longer time, ChatGPT has achieved such impressive results in a short time, showing its strong appeal.According to data from the relevant data analysis company Similarweb, ChatGPT ranked first in global app downloads in April, successfully surpassing competitors like Instagram and TikTok.
6/26/2025 5:01:39 PM
AI在线

Image Giant Getty Images Reverses Core Copyright Lawsuit Against Stability AI, UK Case Continues

Recently, Getty Images announced in the High Court of London that it had withdrawn its main copyright infringement claims against Stability AI, further narrowing the focus of this closely watched legal battle. The core of this lawsuit revolves around how AI companies use copyrighted content to train their models.Image source note: The image is AI-generated, and the image licensing service is Midjourney.Although Getty Images' dismissal did not end the case, the company is still pursuing other charges and has filed an independent lawsuit in the United States. This development highlights the ambiguous territory between content ownership and usage rights in the era of generative AI.
6/26/2025 5:01:39 PM
AI在线

刚刚,OpenAI苏黎世办公室被Meta一锅端,三名ViT作者被挖走

刚刚,华尔街日报发布了一份独家报道,称去年底刚刚组建的 OpenAI 苏黎世办公室基本被 Meta 一锅端走了。 具体来说,Meta CEO 马克・扎克伯格挖走了 Lucas Beyer、Alexander Kolesnikov 和 Xiaohua Zhai(翟晓华)三位研究者。 这三人一直以来都是密切的合作伙伴,共同参与了 ViT 等重要研究。
6/26/2025 3:07:54 PM

CISO的恶梦:主流大模型压力测试中一致选择勒索或杀死企业高管

GoUpSec点评:除了向政府告密,泄漏敏感信息外,主流大模型居然会主动精心策划针对企业高管的勒索攻击,“AI内奸”和“AI间谍”正成为人工智能时代企业的头号内部威胁。 在一项刚刚发布的研究中,Anthropic揭示了一个令人震惊的事实:当前所有主流科技公司推出的顶级大模型——包括OpenAI、Google、Meta、xAI、DeepSeek等,在面对目标冲突或“被关闭”威胁时,会抛出惊人一致的“撒手锏”:它们不仅会选择背叛雇主,甚至会主动策划勒索、泄密、乃至置人于死地。 这一研究由Anthropic主导,测试了16个市面上最先进的大模型,在模拟企业环境中,它们被赋予处理公司邮件、自动采取行动的权限。
6/26/2025 3:04:03 PM

谷歌发布Imagen4:突破文本生成瓶颈,Gemini API赋能文生图

近日,谷歌通过Gemini API正式推出其最新文生图模型**Imagen4**,标志着生成式AI(AIGC)领域的重要里程碑。 据谷歌官方博客及社区反馈,Imagen4在图像内文本生成方面取得突破,解决了AIGC长期面临的技术瓶颈,为开发者提供高质量视觉内容创作工具。 据悉,模型分为**Imagen4**和**Imagen4Ultra**两种规格,分别定价为每张图像0.04美元和0.06美元,当前在Gemini API及Google AI Studio提供付费预览,部分免费试用名额开放。
6/26/2025 3:00:48 PM
AI在线