规范 · 1.3.1
规范
级别是发布者的自我声明,不认证准确性、质量、权利或法律合规性。
规范
判定流程
按顺序回答五个问题。第一个使流程终止的回答会给出级别。
局部
可选。一个数字描述作品的整体。当其中某些部分差别很大时,就把它们分别声明出来。徽章上永远只出现一个数字;细分放在链接背后。
主级别规则。 主级别描述的是承载作品意义的那一部分。理性的读者不会认为对作品具有实质影响的装饰性素材,不决定主级别——把它们声明在这里。
可声明的局部: textimageaudiovideocodedata
JSON
{
"aiUsageScale": "1.0",
"level": 3,
"surfaces": { "text": 3, "image": 5, "audio": 0 }
} 边界情形
下面每一条,都来自试图把决策树问垮的努力。在这里给出定论,是为了让两个诚实的人得到同一个数字。
- 翻译
-
忠实的翻译沿用原作的级别,并附上一条翻译说明。用模型翻译一篇第 1 级的文章,发布出来仍是第 1 级,注明由 AI 翻译。
实质与结构没有改变。惩罚翻译只会让世界上的文字更难被读到,而那恰恰与本标准的初衷背道而驰。
- 转写
-
把你录下的讲话转成文字,属于机械性处理。它不会提高级别。
话是你说的。模型只是把它记了下来。
- 实质来自 AI,文字出自你
-
如果模型想出了实质性决定作品说什么的东西——论点、论证、结构——那它就产出了留在作品里的新材料,即使每个字都是你写的。那是第 2 级或以上,绝不是第 1 级。反馈、批评以及经你核实的事实资料搜集仍是第 1 级:编辑的建议不会让编辑变成作者。
想法就是材料。用自己的话复述并不能取消它们被想出来这件事——而第 1 级的承诺——模型什么也没想出来——必须说到做到。
- 装饰性素材
-
一篇人写的文章配上一张生成的题图,不会让这篇文章变成第 4 级。请在「局部」中声明这张图。
主级别描述的是承载意义的那一部分。否则,一张图库配图就会吞掉整份声明,而人们会干脆不再声明。
- 衍生作品
-
声明你自己那部分贡献的级别;若来源作品已经声明过级别,就把它一并注明。
- 实时生成
-
按需生成、未经审阅就呈现给用户的内容是第 5 级——哪怕系统提示词是由人精心写就的。
- 非生成式机器学习
-
自动对焦、降噪、放大、色彩匹配和分类,凡是不凭空造出任何东西的,都不是生成式 AI。它们不会让你离开第 0 级。
如果你怀疑某个声明
声明是可以公开证伪的。核对链接和元数据;把声明的级别与作品对照,若有出入,再与「局部」细分对照;然后去问作者——在评论里、在评审里、在 issue 里。大多数出入是对边界的诚实解读,而上面的边界情形能解决其中大多数。
如果声明是明知故犯的假话,这套标准的回答是能规模化的那一种:声明曾是公开的、可链接的,如今被公开证伪。这里刻意没有检测器,也没有登记处。声誉就是执行机制——和署名一模一样。
互操作性
与其他词汇表的对应关系仅为建议,不表示 Google、Schema.org、W3C、IPTC 或 C2PA 的认可或支持。
W3C AI Content Disclosure Community Group
The Community Group report approved for publication on 9 October 2026 describes text using human-only, ai-assisted and ai-autonomous. It is not a W3C Recommendation. These are conditional, text-only mapping candidates, not a lossless correspondence: Levels 1 and 2 need additional scope/review information, and the Level 3 candidate requires the review in its full definition. Generic scale snippets omit a CG declaration; assess it independently against the report classification procedure. No endorsement is implied.
IPTC Digital Source Type NewsCodes
Suggested, medium-dependent mappings to the IPTC Digital Source Type vocabulary. The mapping cannot preserve the distinctions between Levels 3, 4 and 5. An AI-assisted text does not automatically qualify as algorithmicallyEnhanced media; choose the IPTC term against the actual asset and its processing history. No IPTC or C2PA endorsement is implied.
C2PA / Content Credentials
Complementary, not competing. C2PA provides tamper-evident, signed provenance about an asset and its processing history. Its core specification does not attribute content to individuals or organisations, and provenance alone cannot determine whose thinking shaped a work. This scale is a declaration about contribution and review, designed to work for web text as well as media files.
EU AI Act (Regulation (EU) 2024/1689), Article 50
Article 50 distinguishes provider marking duties from deployer disclosure duties. Human review or editorial control together with editorial responsibility can affect the public-interest text obligation. A level alone cannot determine legal scope, an exemption, or compliance. Suggested EU icon mappings are illustrative, not official equivalences.
Schema.org and project-defined JSON-LD
DefinedTermSet and DefinedTerm can describe the vocabulary. aiUsageLevel is a project-defined property mapped to https://usagescale.org/spec#aiUsageLevel in an explicit JSON-LD context. The legacy schemaOrgAiUsageLevel mapping field is retained for compatibility; it is not a registered Schema.org property or a supported Google Search feature.
级别映射
| # | Text disclosure candidate | IPTC candidate |
|---|---|---|
| 0 | human-only | digitalCreation |
| 1 | Check scope and review | algorithmicallyEnhanced |
| 2 | Check scope and review | compositeWithTrainedAlgorithmicMedia |
| 3 | ai-assisted | trainedAlgorithmicMedia |
| 4 | ai-assisted | trainedAlgorithmicMedia |
| 5 | ai-autonomous | trainedAlgorithmicMedia |
级别是发布者的自我声明,不认证准确性、质量、权利或法律合规性。
The Community Group's 9 October 2026 report uses
human-only, ai-assisted and ai-autonomous for text.
Levels 1 and 2 do not determine a value without the text's scope and actual review state.
A Level 3 candidate requires the meaningful review in its full definition. Generic scale
snippets omit the separate CG declaration so that publishers can assess it independently.
机器可读形式
权威英文定义集中在一个文件中,并用于生成徽章和机器可读输出。本地化的说明文案另行维护,并接受结构一致性检查。
与其他词汇表的对应关系仅为建议,不表示 Google、Schema.org、W3C、IPTC 或 C2PA 的认可或支持。
GET /zh/levels.json · CC0
HTML
<meta name="ai-usage" content="3">
<meta name="ai-usage-standard" content="https://usagescale.org">
<link rel="ai-disclosure" href="https://usagescale.org/3"> 保留公开定义的链接。如果发布系统支持,可添加自愿提供的元数据。
JSON · /.well-known/ai-disclosure.json
{
"aiUsageScale": "1.0",
"level": 3,
"surfaces": { "text": 3, "image": 0 },
"definition": "https://usagescale.org/3"
} Project-defined JSON-LD property
https://usagescale.org/spec#aiUsageLevel identifies this project's
proposed integer property, with values 0–5. Map it explicitly in a JSON-LD context.
It is not a property in the Schema.org vocabulary, and Google Search has no
documented feature for displaying these levels. additionalProperty
does not have CreativeWork in its published domain and is not the pattern we recommend.
The optional /.well-known/ai-disclosure.json location is also a project convention;
it is not a registered well-known URI or a discovery guarantee. It describes a publisher's
declared scope, not a verified process.
前人的工作
这里没有任何前所未有的东西,假装有,本身就是一种小小的不诚实。
AI Assessment Scale (AIAS)
AIAS describes permitted AI use in assessments. AI Usage Scale describes reported AI use in a completed work. The frameworks serve different purposes.
Creative Commons
The separation between a public mark, a human-readable explanation and machine-readable data informs this project.
Human Provenance in Film
A separate disclosure initiative for film. Consult its own definitions rather than treating its labels as equivalent to these levels.