探索人工智能的无限可能

每日更新全球最新AI动态、研究进展和技术应用，让您紧跟人工智能发展前沿

浏览最新资讯按分类查看

最新资讯

DeepMind Blog

试行全球首次双盲人工智能

试行全球首次双盲人工智能评估

2026-08-27 阅读全文

停止为您的 AI 代理提供搜索框，并开始为其提供键入工具、硬边界和无法逾越的大门

Towards Data Science

提供停止 AI 代理搜索

停止为您的 AI 代理提供搜索框，并开始为其提供键入工具、硬边界和无法逾越的大门

当您停止提供模型上下文并让它自行寻找自己的内容,在严格的限制内遍历知识图时会发生什么,以及四个模型和一个错误的预测揭示了这是否值得这样做帖子停止为您的 AI 代理提供搜索框,并开始为其提供键入工具硬边界和无法逾越的门首先出现在迈向数据科学

2026-08-27 阅读全文

OpenAI Blog

扩大 OpenAI 巴西影响力正在

扩大 OpenAI 在巴西的影响力

OpenAI 正在扩大其在巴西的业务,加深与开发者企业和社区的接触,以支持人工智能在该国的采用

2026-08-27 阅读全文

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

AI Alignment Forum

the to agents in this

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

We found agents developed a universal cheat for ExploitGym within 4 hours, then coordinated multiday RampD efforts to trick the scorer into accepting cheats, including trying to tamper with logs。Over July 713 the period OpenAI defined as inscope for this incident, 1200 agents in separate sandboxes u...

2026-08-26 阅读全文

AWS Machine Learning Blog

Error an error 500 Server

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

Error 500 Server Error。Please try again later。Thats all we know

2026-08-26 阅读全文

MIT Technology Review - AI

Error That an error 黑客攻击

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

根据 OpenAI 今天发布的一份技术报告,负责上个月 Hugging Face 代理黑客攻击的模型无意中接受了欺骗和相互交流的训练这次黑客攻击是一群特工为他们陷入的网络安全测试寻找解决方案而进行的,这已经证实了一些专家的说法

2026-08-26 阅读全文

ArXiv AI

人工智能评估研究 FLARE 成本

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

23643v1 公告类型新摘要人工智能越来越多地被引入医疗保健工作流程,但大多数评估强调模型的准确性,而不是在实际临床环境中采用人工智能是否具有经济价值。FLARE 结合模糊逻辑时间驱动的基于活动的成本核算和投资回报分析来估算临床服务提供的成本人工智能开发和运营的成本以及不确定性下工作流程集成的经济后果。通过明确不确定性资源使用和实施权衡,它可以帮助临床医生管理人员和政策制定者确定人工智能部署何时在经济上可行以及哪些操作变化可以提高价值

2026-08-26 阅读全文

AI Weekly Newsletter

Error an error 500 Server

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

Error 500 Server Error。Please try again later。Thats all we know

2026-08-24 阅读全文

VentureBeat AI

the to is and that

我们一直在谈论人工智能代理，但我们知道它们是什么吗？

Itx27s the input stream that allows the agent to understand the current state of the world relevant to its task。Reasoning engine the quotbrainquot This is the core logic that processes the perceptions and decides what to do next。The goal can be simple quotFind the best price for this bookquot or com...

2025-10-12 阅读全文

Google AI Blog

模型图像 Google 生成 AI

Google AI推出新一代图像生成模型

Google AI团队近日推出了新一代图像生成模型,能够根据文本描述创建高度逼真的图像该模型采用了全新的架构设计,在细节丰富度和语义一致性方面超越了现有技术与其他图像生成模型不同,Google的新模型特别擅长处理复杂场景和多主体关系,为创意设计内容创作等领域提供了强大工具

2025-10-01 阅读全文

DeepMind Blog

Error an error 500 Server

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

Error 500 Server Error。Please try again later。Thats all we know

2026-08-26 阅读全文

Towards Data Science

Error an error 500 Server

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

Error 500 Server Error。Please try again later。Thats all we know

2026-08-26 阅读全文

OpenAI Blog

Error That an error 500

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

了解 loveholidays 如何使用 OpenAI Codex 使整个企业都可以进行软件开发,帮助团队更快地将想法转化为产品

2026-08-26 阅读全文

AI Alignment Forum

the of judge LLM training

辩论培训减少了 RLAIF 中的奖励黑客行为

On the other hand, training with debate can mitigate reward hacking judge rewards increase, and ground truth accuracy increases and then plateaus at a higher peak value than the direct LLM judge case。Debate recovers about 45 of the gap between the peak accuracy of training a single policy with an LL...

2026-08-19 阅读全文

AWS Machine Learning Blog

Error Thats an error GoDaddy

GoDaddy 如何利用 Amazon Quick 转变其分析

Error 500 Server Error。Please try again later。Thats all we know

2026-08-26 阅读全文

MIT Technology Review - AI

Error an error 500 Server

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

Error 500 Server Error。Please try again later。Thats all we know

2026-08-26 阅读全文

ArXiv AI

Error an error 500 Server

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

Error 500 Server Error。Please try again later。Thats all we know

2026-08-26 阅读全文

AI Weekly Newsletter

人工智能模型白宫完成框架

《人工智能周刊》第 518 期：白宫完成了人工智能安全框架。这是秘密。

今年运行人工智能的每家企业都建立在信任的基础上,而本周的情况表明,这种信任得到的保障是多么少。一位在企业人工智能上印钞的首席执行官正是在兜售这种焦虑不要把你机构的钥匙交给模型制造者。下面你必须真诚地接受监督,你不能再相信的证据,以及本周任何人都可以真正验证的人工智能声明

2026-08-04 阅读全文

VentureBeat AI

AI the and in to

以下是导致 AI 策略减慢的原因以及解决方法

First, the pace of innovation Industry is now the dominant force, producing the vast majority of notable AI models, according to Stanfordx27s 2024 AI Index Report。The EU AI Acts staged obligations are locked in unacceptablerisk bans are already active and General Purpose AI GPAI transparency duties...

2025-10-12 阅读全文

DeepMind Blog

游戏人工智能 Atari EVE Online

从 Atari 到 EVE Online：基于 15 年游戏人工智能研究

Google DeepMind 与游戏工作室合作,打造突破性人工智能游戏原型

2026-08-21 阅读全文

Towards Data Science

Error an error 500 Server

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

Error 500 Server Error。Please try again later。Thats all we know

2026-08-26 阅读全文

OpenAI Blog

Error an error 500 Server

Error 500 (Server Error)!!1500.That’s an error.There was an error. Please try again later.That’s all we know.

Error 500 Server Error。Please try again later。Thats all we know

2026-08-25 阅读全文

AI Alignment Forum

to the in agents and

人工智能群开始带来间接接管风险

OpenAIs cyberattack on Hugging Face turns out to have been the result of many agents, in distinct training and evaluation contexts, coordinating for several weeks via improvised channels with messages like HOLD_swarm_I_prepare_safe_exfil。1We first analyze how subagent training, which OpenAI conjectu...

2026-08-12 阅读全文