欧盟 AI Omnibus 法案生效，高风险 AI 规则适用推迟至 2027 年 12 月
entered into force on 27 July 2026
- 1欧盟 AI Omnibus 于 2026 年 7 月 27 日生效，对 AI 监管规则定向简化，保留安全与基本权利保障
- 2附件三高风险 AI 规则 2027 年 12 月 2 日起适用，嵌入机械、玩具等实体产品的高风险 AI 推迟至 2028 年 8 月 2 日
- 3SMEs 专属简化义务扩展至小型中盘企业（SMCs），监管沙盒扩容并新设欧盟级沙盒，AI 素养要求简化
- 4新规禁止生成非自愿性内容或儿童性虐待材料的 AI，AI Office 对嵌入大型平台和搜索引擎的系统扩大监管
欧盟给高风险 AI 合规大限松绑、减轻中小企业义务，监管底线并未撤除，是一次为创新让路的定向简化。
- 影响：做欧盟市场高风险 AI 业务的企业合规窗口延长，排期与预算可按新节点重算；中小规模企业义务简化并可用监管沙盒降低试错成本；生成亲密内容类 AI 被划入红线，平台内基于通用模型的系统面临 AI Office 更多监管。
- 参考：可考虑按 2027 年 12 月与 2028 年 8 月两个节点重排欧盟高风险 AI 合规项目；符合 SMC 条件的在欧主体值得核对能否套用简化义务；内容生成类产品线先自查是否触及新禁令。
- 依据：欧盟委员会官方公告，生效日期、推迟时间表与各项简化措施均为官方口径，执行细节以其立法文本与 FAQ 为准。
On 27 July 2026, the AI Omnibus enters into force across the EU, bringing extended timelines to administrative simplification.
The AI Omnibus, proposed as part of the digital omnibus package on 19 November 2025, delivers a targeted simplification of the AI rulebook while preserving strong safeguards for people’s safety and fundamental rights.
It supports innovation and competitiveness by easing compliance for smaller businesses, extending timelines, expanding testing and experimentation opportunities, and providing greater legal clarity for companies developing and deploying AI in Europe.
The updates introduce:
Innovation support
- Support for growing companies: Some measures previously reserved for SMEs are extended to small mid-cap companies (SMCs), helping more businesses benefit from a proportionate approach
- Testing and experimentation: Expands access to regulatory sandboxes and introduces an EU-level regulatory sandbox, giving companies more space to develop and test AI systems under regulatory supervision
Extended timelines
- High-risk AI systems in Annex III: Rules apply starting 2 December 2027
- High-risk AI embedded in physical products (machinery, toys, lifts, etc.) in Annex I: Rules apply starting 2 August 2028
Reduced administrative burdens
- Simplification for SMCs: Extends simplified legal obligations for SMEs to small and mid-cap companies
- AI literacy: Previous AI literacy requirement for companies is simplified, with the Commission and the Member States taking a stronger role in promoting AI literacy
- EU database registration: Simplifies obligation to register exempted AI systems in the EU central database
Safety and fundamental rights
- Ban on nudification apps: Prohibits AI systems that generate non-consensual sexually explicit and intimate content or child sexual abuse material
- Bias detection: Allows for the processing of special categories of personal data to detect and correct bias
AI governance alignment
- Extended enforcement powers: The AI Office will have extended oversight of certain AI systems, including those built on general-purpose models and embedded in large online platforms and search engines
- Legal and procedural clarity: Clarifies interplay between the AI Act and other EU laws and simplifies procedures for conformity assessment bodies
Resources
Access the full legislative text
来源：欧盟 AI Act 官方 · digital-strategy.ec.europa.eu