EU AI Act
PulseAugur coverage of EU AI Act — every cluster mentioning EU AI Act across labs, papers, and developer communities, ranked by signal.
- developed by European Commission 95%
- instance of European Union 90%
- affiliated with European Union 70%
- affiliated with European Parliament 70%
- affiliated with European Commission 70%
- instance of General Data Protection Regulation 70%
- used by arXiv 60%
- instance of Health Insurance Portability and Accountability Act 60%
- affiliated with Mastodon 50%
- 2026-05-25 regulatory The EU AI Act's compliance obligations are shifting focus to supply chains, with information requests preceding official deadlines. 来源
- 2026-05-18 regulatory The EU AI Act begins enforcement in 75 days. 来源
- 2026-05-08 regulatory EU lawmakers are concerned that current cybersecurity laws are insufficient to address advanced AI hacking tools like Anthropic's Mythos.
17 天有情绪数据
-
AI初创公司推出治理、品牌分析和公关工具
多家AI公司推出了旨在改进AI开发和应用各个方面的新产品。Premetric.ai发布了一个用于部署前AI治理的平台，帮助组织在构建AI系统前进行评估，并符合欧盟AI法案等标准。Waikay推出了“Topical Presence”，一种用于理解AI如何感知品牌主题关联性的指标。此外，Barney、Dappit和Agent1等新的免费AI代理工具正在涌现，以简化去中心化开发，而Newsworthy.ai则推出了一个AI驱动的新闻稿平台…
-
本地AI工具助力欧盟AI法案下的行政表格处理
Behördenhelfer v2 是一款新推出的开源工具，旨在协助处理行政表格，完全在本地运行，无需依赖云端。它支持多语言，并与LibreOffice集成，使用Ollama进行推理。该工具在设计时考虑了欧盟AI法案的原则，侧重于结构化工作流程和人工监督，而非开放式聊天机器人交互。
-
EU AI Act misses core issue: AI literacy needs governance, not just model training
Europe's EU AI Act mandates AI literacy for employees, but the focus on model behavior and prompt engineering misses the practical application gap. The author argues that AI systems are retrieval engines, not sources of…
-
Golem offers 15% discount on AI workshops through May
Golem Karrierewelt is providing a 15% discount on its AI workshops and e-learning programs. These courses cover topics such as Microsoft Copilot, fundamental AI concepts, and the EU AI Act. The promotion is valid until …
-
Anthropic 的 Claude 4.7、Qwen Image 2.0 和无服务器 GPU 亮点
本次 TLDR AI 新闻通讯涵盖多项 AI 发展，包括 Anthropic 的 Claude 4.7 模型、阿里巴巴的 Qwen Image 2.0 以及无服务器 GPU 的进展。它还推广了 SANS 关于 AI 安全成熟模型的一本电子书。
-
AI Security Summit in London to focus on EU AI Act compliance
An AI Security Summit is scheduled for May 14th in London, focusing on critical security and governance challenges for AI engineers. Organized by Snyk, the event will address compliance with the EU AI Act and emphasize …
-
AI事件框架提出8项国际响应标准
一项新框架提出了八项标准，用于确定何时AI事件需要国际响应。该框架旨在标准化升级流程，确保及时进行跨境协调，以遏制和减轻AI风险。它涵盖了操纵、失控和CBRN威胁等关键领域，并针对现实世界事件进行了测试。研究还指出了现有框架（如欧盟AI法案）中可能存在的低估问题。
-
欧盟委员会澄清AI规则以打击数字欺骗
欧盟委员会已提议新法规，旨在提高AI的透明度并打击数字欺骗。这些规则特别针对AI生成的内容，要求清晰的标签以防止用户被误导。该举措是更广泛的欧盟AI法案的一部分，侧重于生物识别数据和面部识别等领域，以确保AI的道德部署。
-
Bizsuite推出Air用于AI代理审计追踪，简化合规性
Bizsuite推出了一款名为Air的开源工具，旨在为AI代理提供防篡改的审计追踪。该工具将SOC2、ISO 27001和欧盟AI法案中的22项控制措施进行了映射。虽然Air负责安全日志记录的技术实现，但Bizsuite专注于将这些技术细节转化为审计师和采购团队易于理解的摘要，他们声称这一过程可以在四小时内完成。
-
Anthropic Claude Platform now available natively via AWS
Anthropic's Claude Platform is now accessible directly through AWS accounts, simplifying integration for developers already using the cloud provider's services. This move expands Anthropic's enterprise reach and offers …
-
AWS MCP 服务器的推出造成审计合规性差距
AWS 推出了其 MCP 服务器，使编码代理能够通过 IAM 范围的权限访问 AWS 控制平面。此功能可显著提高生产力，但也带来了合规性挑战，尤其是在审计日志记录要求方面。现已推出一款新的“bizsuite ai-audit kit”，售价 997 美元，用于解决这些日志记录差距，确保代理的操作、输入和输出被记录下来并且防篡改，以便审计员查阅，尽管欧盟人工智能法案的截止日期已延长。
-
欧盟委员会就人工智能透明度指导方针草案进行咨询
欧盟委员会已启动一项关于人工智能透明度要求的指导方针草案的公众咨询。这些指导方针旨在明确人工智能系统的义务，特别是那些根据《欧盟人工智能法案》被指定为高风险的系统。此次咨询旨在收集利益相关者的意见，以便在最终确定前完善拟议规则。
-
AI literacy becomes boardroom priority, impacting employee pay and investment
Companies are increasingly prioritizing AI literacy among their workforce, driven by market demand and tightening regulations like the EU AI Act. Investors view AI literacy as a key indicator of a company's future succe…
-
欧盟人工智能法案要求当前的AI控制，但方法需要改进
欧盟人工智能法案已强制规定了所讨论的控制类型，尽管目前的方法尚未足够完善。这暗示着，虽然监管框架已经存在，但执行的实际实施和技术能力还需要进一步发展。
-
欧盟 AI 标准 prEN 18228 未能解决决策问题
一项新的欧洲标准 prEN 18228，旨在通过要求组织在 AI 系统的整个生命周期中识别危害、评估风险和监控控制措施来正式化 AI 风险评估。虽然该标准带来了结构和产品安全理念，但它可能无法充分解决 AI 系统的独特故障模式，这些模式可能依赖于上下文，并且仅在部署后才能显现。该标准依赖于传统的概率乘以严重性模型，可能会模糊频繁的低影响错误与罕见的严重故障之间的关键区别，从而可能导致在系统部署方面的决策不足。
-
New benchmark models strategic auditee gaming under continuous compliance monitoring
Researchers have developed a new benchmark and formalization for strategic gaming in continuous compliance monitoring, particularly relevant for regulations like the EU AI Act. The study models this as a T-round Stackel…
-
Paper proposes ethical AI framework for driver monitoring systems in AVs
A new research paper proposes an ethical design framework for multimodal driver monitoring systems (DMS) in automated vehicles. The paper identifies gaps in current regulations like GDPR and the EU AI Act concerning in-…
-
PCIe 8.0 draft targets 1 TB/s, AMD launches new AI GPU, EU delays AI Act rules
The PCI-SIG has released a draft specification for PCIe 8.0, aiming to double the data transfer rate to 1 TB/s per lane. This advancement is crucial for supporting the increasing demands of AI and high-performance compu…
-
ServiceNow enhances AI Control Tower with new discovery, monitoring, and governance features
ServiceNow has announced significant updates to its AI Control Tower platform, enhancing the discovery, monitoring, and governance of AI systems across cloud and enterprise applications. The update introduces thirty new…
-
意大利总理梅洛尼在AI内衣照疯传后谴责AI深度伪造
意大利总理乔治亚·梅洛尼公开反对AI生成的深度伪造图片泛滥，其中包括一张疯传的她穿着内衣的图片。她强调，虽然她可以保护自己，但许多人却不能，并敦促公众在分享信息前进行核实。梅洛尼政府一直在积极监管AI，此前已通过一项全面的法律，对深度伪造等有害AI应用处以监禁。