报道称 Anthropic 或为 Claude 输出增加可检测标记，核心部署细节尚待证实

to disclosing the existence of the generated or manipulated content “in an appropriate manner that does not hamper the display or enjoyment of the work”. [...] Providers of AI systems (including general-purpose AI systems) that generate synthetic audio, image, video, or text must ensure that outputs are both marked in a machine-readable format and detectable as AI-generated. This is a technical obligation aimed at provenance: enabling detection tools to verify whether content is AI-generated. The specific technical standards for this marking are being developed through the Code of Practice and complementary EU standardisation work. Providers should track these developments closely, as practical implementation details are still being finalised ahead of August 2026. [...] While these obligations apply from 2 August 2026, the AI Omnibus provisional agreement of May 2026 grants generative AI systems already on the market before that date until 2 December 2026 to meet the machine-readable

signaling) that has been enabled and tested rather than merely available in the underlying model or platform. [...] ### Short-Term Mitigations [...] ### Strategic Considerations

2 August 2026. Article 50 of the EU AI Act becomes enforceable. The European Commission published the second draft Code of Practice on marking and labelling of AI-generated content on 5 March. The final Code lands May or June. That gives product teams roughly eight weeks between final text and enforcement. The Code is technically voluntary. In practice, it is the benchmark regulators will use. If you are a provider, you build or deploy a generative AI system producing audio, image, video, or text: - Your outputs must carry at least two layers of machine-readable active marking - Layer 1: digitally signed, tamper-evident metadata - Layer 2: an imperceptible watermark interwoven in the content, designed to survive compression, cropping, and format changes [...] Meanwhile, Reuters reported that within the federal government, Anthropic’s models were widely viewed as more capable than rival offerings. Reuters also reported one contractor saying Claude was “the best,” while Grok often gave i

### Mark AI-generated content so it can be detected (Article 50(2) - providers) Providers of AI systems that generate synthetic audio, images, video or text must: (i) mark their outputs in a machine-readable format; and (ii) provide a corresponding means of detection – so that people can identify these as artificially generated or manipulated. Marking alone, without a detection solution, is not sufficient. The marking must be embedded using metadata, watermarks, or other interoperable techniques - not visible to end users, but effective and robust enough to survive routine modifications. [...] Skip to Main Content Skip to Footer Travers Smith Legal briefing | | # Is it a bot? EU AI Act transparency rules take effect 2 August 2026 Is it a bot? EU AI Act transparency rules take effect 2 August 2026 ## Overview [...] 7. Update governance and document exemptions. If you rely on any exception (e.g., obviousness, editorial responsibility, artistic use), document the reasoning. Bus

#### 优设提醒：您的赞赏金额将直接打赏给作者 ### 微信赞赏 ## 好内容值得被赞赏 #### 学到了，来请作者喝杯咖啡吧 数字生命卡兹克 #### 优设提醒：您的赞赏金额将直接打赏给作者 [...] AI 智能体电商设计工作流特训营（第9期） #### 10年+大厂设计专家出品 ##### 爆款热卖 ## 零基础 Codex 教你做第1款游戏极速入门指南！ #### 教你从零开始做出一款游戏！ ## 15节零基础 Suno AI 音乐全流程实战必修课 #### 掌握最新的AI音乐生成全流程 ### 学教程 #### 3000+免费教程在线观看 ### 找软件 #### 省时高效的设计神器都在这 ### 知识树 #### 热门软件零基础入门 ### 字体下载 #### 最新免费可商用字体 ## 搜索 ChatGPTFable 5CodexFigmaPrompt即梦GeminiGPT Image 2AIGCOpenClawNano Banana优设标题黑作品集opc一人公司GPT私单版权海报设计设计规范可商用字体 找免费字体？看这里 热门！超火的300+AI工具 重磅！AI星踪岛上线 推荐！专业创作Skill社区 2026零基础必学好课 ### 近期热门 #### 6步搞定！用AI Skill把平面图变成爆款海报 ##### 2.1w 设计师在看 #### 我把 Logo 设计师的工作，全交给 AI 了！ ##### 2.3w 设计师在看 #### 5分钟学会！GPT-image-2电商&海报提示词实操教程 ##### 2.8w 设计师在看 #### 1分钟学会！用GPT把海报拆成Photoshop分层文件 ##### 4.3w 设计师在看 #### 灵感不设限！10 个值得收藏的设计灵感和工具网站 ##### 3.2w 设计师在看 ### 微信赞赏 ## 好内容值得被赞赏 #### 学到了，来请作者喝杯咖啡吧 #### 优设提醒：您的赞赏金额将直接打赏给作者 ### 微信赞赏 ## 好内容值得被赞赏

79 comments ### Transcript: [...] points towards something more scalable than trying to enumerate and patch every individual failure mode. All right, drop your questions in the comments, subscribe if you want to stay ahead of what is coming next in AI. Thanks for watching, and I'll catch you in the next one. [...] So, Anthropic had a genuinely big week, and there's a lot to get through. Some of it exciting, some of it raising real questions, and at least one story from OpenAI that's worth your attention if you care about where alignment research is actually heading. Let's start with the announcement. Anthropic launched Claude Tag, positioned as an evolution of Claude Code, but redesigned for entire teams rather than individual developers. Anthropic's headline stat is that around 65% of the company's own product code is now being written with Claude Tag's involvement. Andrej Karpathy, who joined Anthropic not long ago, framed [music] it as the third major transformation of LLM interface