Google DeepMind
Google DeepMind is one of the entities PulseAugur tracks across the AI industry. This page surfaces every recent cluster mentioning Google DeepMind — vendor announcements, third-party press, social commentary, research papers, and regulatory filings — ranked by signal across our 200+ source set. Linked to the canonical entity record on Wikipedia and Wikidata so the entity card AI engines build is grounded in the same identity Wikipedia uses, not a slug-collision lookalike.
- founded by Mustafa Süleyman 100%
- subsidiary of Google+ 100%
- subsidiary of GOOGL 100%
- subsidiary of Alphabet Inc. 100%
- developed Gemini 2.5 Pro 95%
- affiliated with Google Research 95%
- developed by Gemini Omni 95%
- developed by Google+ 95%
- instance of Gemini Omni 95%
- spun off from Isomorphic Labs 95%
- developed AI Pointer 95%
- developed Genie 3 95%
- 2026-05-25 research_milestone Google DeepMind's AlphaProof Nexus AI solved nine open Erdős mathematics problems. 来源
- 2026-05-24 research_milestone An AI agent developed by Google DeepMind autonomously solved 9 open mathematical problems. 来源
- 2026-05-23 partnership Google DeepMind is expanding its partnership with Singapore to focus on the safe deployment of AI. 来源
- 2026-05-22 product_launch Google DeepMind announced the expansion of its SynthID watermarking technology and new AI detection features in Gemini and Google Search.
- 2026-05-21 product_launch Google DeepMind launched the Google DeepMind Accelerator program in Asia Pacific. 来源
- 2026-05-20 research_milestone Demis Hassabis discussed the potential of AI, including Gemini for Science and AlphaGenome, to solve diseases at Google I/O.
- 2026-05-20 controversy An AI engineer filed a legal claim against Google DeepMind alleging unfair dismissal after protesting the company's work for the Israeli government. 来源
- 2026-05-19 research_milestone Google DeepMind CEO Demis Hassabis described the current AI advancements as a 'profound moment for humanity' and the 'foothills of the singularity' during the Google I/O keynote.
- 2026-05-19 product_launch Google DeepMind released three new experimental AI-powered tools to assist scientists. 来源
- 2026-05-17 hiring Google DeepMind UK workers voted to unionize. 来源
- 2026-05-17 product_launch Google DeepMind launched the "AI for the Planet" accelerator program in the Asia Pacific region. 来源
- 2026-05-12 product_launch Google DeepMind introduces an experimental AI-enabled mouse pointer. 来源
- 2026-05-12 product_launch Google DeepMind introduces an experimental AI-enabled mouse pointer. 来源
- 2026-05-12 product_launch Google DeepMind revealed a concept for an AI-powered adaptive mouse pointer. 来源
- 2026-04-24 research_milestone Google DeepMind CEO Demis Hassabis declared humanity is in the 'foothills of the singularity' at Google I/O. 来源
20 天有情绪数据
-
AI safety research faces sabotage risk as auditors fail to detect flaws
Researchers have developed a new benchmark called Auditing Sabotage Bench to test the ability of AI models and humans to detect subtle sabotage in machine learning research codebases. The benchmark includes nine ML code…
-
Google DeepMind推出AI联合临床医生，助力医护人员和患者
Google DeepMind推出了一项AI联合临床医生研究计划，旨在增强医护人员的能力并改善患者护理。该计划建立在MedPaLM和AMIE等先前工作的基础上，探索AI如何作为协作团队成员，在临床监督下与患者互动。早期评估显示，AI联合临床医生在药物知识和推理方面表现出色，优于现有的证据综合工具。
-
Economist urges caution on AI job impact; UAE bets big on AI infrastructure
Nobel laureate economist Daron Acemoglu maintains a cautious stance on AI's impact on employment, arguing that AI agents are more likely to augment specific tasks rather than replace entire jobs due to the complexity an…
-
前Google DeepMind研究员创立的初创公司Ineffable Intelligence融资11亿美元
由前Google DeepMind研究员创立的AI初创公司Ineffable Intelligence已获得创纪录的11亿美元种子轮融资。该公司专注于超级智能的开发，投后估值达到510亿美元。本轮融资由Sequoia Capital和Lightspeed Venture Partners共同领投，Nvidia、DST Global、Index Ventures、Google和英国国家AI主权基金等知名机构也参与其中。
-
Google员工敦促CEO因道德原因拒绝五角大楼AI合同
包括DeepMind和Google Cloud的员工在内的600多名Google员工签署了一封致CEO Sundar Pichai的信。他们敦促公司拒绝为五角大楼提供机密工作。员工们引用了对致命自主武器和大规模监控的担忧。
-
Google DeepMind研究员质疑AI意识；美国支持xAI的偏见法
一位Google DeepMind的研究员发表了一篇论文，认为由于他称之为“抽象谬误”，AI永远无法实现意识。另外，美国司法部已介入xAI针对科罗拉多州AI偏见法提起的诉讼。司法部认为，该州法律SB24-205迫使AI开发者采取歧视性做法，并违反了平等保护条款。
-
Fireworks AI 将 Google 的 Gemma 4 模型添加到其训练平台
Fireworks AI 宣布已将其训练平台整合了 Google DeepMind 的 Gemma 4 模型，特别是 26B 和 31B 参数版本。此次整合允许用户利用 Fireworks 的托管和训练 API 工作流对这些模型进行微调。该平台支持监督微调 (SFT) 和直接偏好优化 (DPO)，并提供可自定义的损失函数和 256K 上下文窗口，预计很快将支持强化学习 (RL)。
-
Google DeepMind paper argues LLMs will never achieve consciousness
A paper from Google DeepMind researchers posits that artificial intelligence, including large language models, will never achieve consciousness. The argument centers on the idea that current computational architectures …
-
DeepMind founder David Silver raises $1.1B for AI that learns without human data
Ineffable Intelligence, a new AI lab founded by former DeepMind researcher David Silver, has secured $1.1 billion in funding. The company aims to develop a "superlearner" that can acquire knowledge and skills autonomous…
-
Google DeepMind 讨论 AI 在韩国的作用；新的 Python 框架助力角色动画
一个名为 ai4animationpy 的新 Python 框架已被推出，旨在利用神经网络生成和控制角色动作。该工具旨在增强 AI 领域的动画制作和创意工作流。另外，Google DeepMind 正在与韩国政府讨论 AI 如何加速科学发现和促进区域经济增长，这标志着 AlphaGo 问世十周年。
-
AI在工作场所的经济可行性仍不确定，研究表明模型在处理现实世界任务时遇到困难。
尽管AI能力取得了重大进展，但许多系统在现实世界的工作场所应用中仍难以证明其经济可行性。Mercor的一项最新研究发现，来自OpenAI、Anthropic和Google DeepMind等领先公司的AI代理未能完成最常见的专业任务。这一差距凸显了将AI整合到现有工作流程中的挑战，以及在实现变革性业务影响之前，需要超越基本编码熟练度的进一步发展。
-
DeepMind的AlphaGo负责人David Silver推出Ineffable Intelligence，获得Sequoia投资
David Silver，DeepMind的AlphaGo及其他AI项目的关键人物，已成立一家名为Ineffable Intelligence的新研究实验室。该实验室旨在创建一个“超级学习者”，通过直接经验而非现有数据的预训练来获取知识。这种根植于强化学习的方法，旨在使AI能够在科学和数学领域做出新颖的发现。
-
AlphaGo 创造者 David Silver 为不同于 LLM 的 AI 方法筹集 11 亿美元
AlphaGo 背后的 AI 研究员 David Silver 创立了 Ineffable Intelligence 公司，旨在探索通往超级智能的不同路径。他的公司将专注于强化学习，即 AI 通过试错进行学习，这与大多数实验室偏爱的大型语言模型 (LLM) 方法形成对比。Silver 认为，从人类数据中学习的 LLM 最终将无法实现真正的超级智能，而能够独立从世界中学习的系统则可以。Ineffable Intelligence 已获得…
-
Google DeepMind 科学家认为人工智能永远无法实现意识
Google DeepMind 科学家 Alexander Lerchner 发表了一篇论文，认为大型语言模型（LLM）和其他计算系统永远无法实现意识。他的论点被称为“抽象谬误”，认为人工智能只能模拟有意识的行为，因为它缺乏内在意义，并且需要人类的解释来组织数据。这一观点与人工智能公司领导者经常做出的乐观的通用人工智能（AGI）预测形成对比，暗示了人工智能的实际和商业能力的潜在上限。
-
AI weather forecasts may miss rare, extreme events, experts warn
While AI models are rapidly improving weather forecasting accuracy and efficiency, experts express concern about their reliability in predicting unprecedented "gray swan" events. These rare but plausible extremes, exace…
-
Local LLMs challenge ChatGPT, Claude, and Gemini as AI shifts power dynamics
The discussion around AI models like ChatGPT and Claude highlights a shift in skill distribution, where those with financial resources can leverage the trained abilities of others, potentially marginalizing the original…
-
Google DeepMind 与韩国合作，推动人工智能驱动的科学发现
Google DeepMind 宣布与大韩民国建立新的合作伙伴关系，旨在利用先进的人工智能实现科学突破和创新。此次合作将在首尔建立一个人工智能园区，让韩国研究人员能够使用 Google 的前沿人工智能模型和专业知识。该计划侧重于在生命科学、天气和气候等领域加速发现，并继承了十年前 AlphaGo 带来的影响。
-
AI robots now beat elite table tennis players, showcasing advanced robotics
Researchers have developed an AI-powered robot capable of competing at a high level in table tennis. This robot has demonstrated a significant win rate against human players, even those considered elite. The system's ab…
-
Google DeepMind 的 Vision Banana 统一了 AI 生成与感知
Google DeepMind 的研究人员开发了 Vision Banana，这是一个基于 Nano Banana Pro 构建的模型，通过将图像转换为其他图像来处理视觉任务。这种方法迫使模型生成像素，从而赋予其对 3D 几何和深度的理解。因此，与专用模型相比，Vision Banana 在零样本分割和深度估计方面表现出卓越的性能。
-
Google DeepMind衍生公司Isomorphic Labs为AI药物试验筹集21亿美元
Isomorphic Labs是一家由Google DeepMind孵化的人工智能驱动的药物发现公司，已在B轮融资中获得超过20亿美元。这笔巨额投资由Thrive Capital领投，将用于扩展其专有的药物设计引擎IsoDDE，并将其候选药物管线推进到人体临床试验阶段。该公司利用DeepMind的AlphaFold技术来预测蛋白质结构并设计高效、低副作用的药物。