SB 1000 AI Disclosure Verifier

SKILL效率
效率 未标注权限 许可证不明确
安装这个技能

加州 SB 1000(California AI Transparency Act,2026-09-30 生效)AI 披露标识核验:检查 AI 生成的图像/视频/音频是否包含美国加州法定的六项潜在披露字段(§ 22757.3(a)(1)),逐字段报告 present/absent 并给出匹配证据。六字段为:提供者名称 · 系统名称与版本 · 时间与日期 · 唯一标识符 · 系统是创建还是修改了内容 · (2029-01-01 起)是否主要作为辅助技术。法条允许「直接嵌入」或「链接到永久网页」两种方式。触发词:加州AI法案、加州AI透明度法案、California AI Transparency Act、SB 1000、SB1000、22757、AI披露、AI标识核验、latent disclosure、unique identifier、digitalSourceType、C2PA、Content Credentials、content provenance、深度伪造披露、机器可读标记、EU AI Act 第50条、Article 50、透明度义务、跨境合规、AI 内容合规、生成合成内容标识、发布前自查、合规证据报告。Verify whether AI-generated media carries the six latent-disclosure fields required by California SB 1000 (effective 2026-09-30): provider name, system name and version, time and date, unique identifier, created-or-altered flag, and assistive-technology status from 2029. Reports per-field status with matched evidence. [Compliance window] EU AI Act Art.50(2) machine-readable marking: systems generating synthetic audio, image, video or text placed on the market before 2026-08-02 must be compliant by **2026-12-02**.

一句话选型结论

按资料字段整理

SB 1000 AI Disclosure Verifier是效率类的 Agent Skill,安装方式未标注,权限级别未标注,许可证不明确,有可直接复制的安装命令。技能本质是给模型读的说明书,效果取决于模型与客户端;装完先用一个小任务验证,再放进正式流程。

分类效率 安装方式未标注 权限级别未标注权限 许可证许可证不明确
累计安装 未标注 本周 未标注
GitHub 星标 未标注 仓库 未标注
下载量 未标注 技能市场下载数
权限级别 未标注权限 安装前请确认风险范围
许可证 许可证不明确 未说明

安装

未标注
在支持 Skills 的客户端里执行,装完问一句「你有哪些技能」验证
skillhub install user_333cb607/sb1000-ai-disclosure-verifier
  1. 1
    确认客户端支持

    确认你的客户端支持 Skills(Claude、Codex、Cursor 等),并升级到较新的版本。

  2. 2
    拿到技能文件

    执行上面的命令安装;也可以把仓库里的 SKILL.md 手动放进客户端的 skills 目录。

  3. 3
    刷新技能列表

    重新打开客户端,或刷新技能列表,让新放进来的技能被加载。

  4. 4
    验证是否生效

    问一句「你有哪些技能」,确认列表里出现它,再拿一个小任务试跑。

技能档案

效率
字段来自技能仓库与公开索引,中文说明由本站标注
作者 / 维护方user_333cb607仓库 未标注 · 版本 未标注
宿主客户端未标注技能可被这些客户端加载
安装方式未标注安装包 community
权限级别未标注权限安装前请确认风险范围
许可证未说明许可证不明确,商用前请自行确认条款
仓库可用性available技能仓库当前是否可访问
最近核验未标注上游最后核验时间 · 数据更新 1791737220
正文体量9.7 KBSKILL.md 原文大小,越大说明说明书写得越细

典型使用场景

3 类
按分类整理的常见用法,具体以 SKILL.md 为准
整理待办与日程

把零散任务收进清单,排出优先级。

批量处理重复任务

把固定动作做成模板,重复执行。

把流程写成清单

把口头流程整理成可复用的步骤。

选择建议

装之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 有可直接复制的安装命令,不用自己翻文档找入口。

先想清楚的地方

  • 没有标注权限级别,实际能做什么要装好后确认。
  • 许可证不明确,商用或二次分发前先到仓库确认条款。
  • 没有做站内核验,内容以作者仓库为准。
  • 技能只是说明书,实际效果取决于模型与客户端,装完先用小任务试一遍。

更适合

  • 需要在效率场景里给模型补一套做法的人。
  • 愿意先读一遍 SKILL.md、确认它会让模型做什么的人。

不太适合

  • 对许可证要求很严、必须明确可商用的人。
  • 只想装完就自动变强、不打算看说明书的人。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
技能权限级别安装方式累计安装宿主客户端安装命令
SB 1000 AI Disclosure Verifier当前 未标注权限 未标注 未标注 未标注 有
Nano PDF 文档 联网访问 ClawHub 市场 424.9 万 OpenClaw 有
自动更新程序 浏览器控制 ClawHub 市场 296.4 万 OpenClaw 有
Gemini 助手 联网访问 ClawHub 市场 254.7 万 OpenClaw 有
word DOCX 文档 联网访问 ClawHub 市场 239.6 万 OpenClaw 有
Excel XLSX 前端 联网访问 ClawHub 市场 227.0 万 OpenClaw 有

SKILL.md 全文

9.7 KB
模型实际读到的说明书原文,来自技能仓库

SB 1000 AI Disclosure Field Verifier

> Compliance window (countdown): EU AI Act Art.50(2) requires
> machine-readable marking. Systems generating synthetic audio, image, video or text
> placed on the market before 2026-08-02 must be compliant by 2026-12-02.
> That date is from the official implementation timeline, not marketing copy.

California SB 1000 (Chapter 861, Statutes of 2026), which amends the California AI
Transparency Act, was approved and filed on 30 September 2026 and took effect
immediately as an urgency statute. It amends Business and Professions Code
sections 22757.1 through 22757.5 and adds section 22757.4.1.

This Skill answers one narrow question:

> Are the six statutory latent-disclosure fields actually present in this file?

It is a field-completeness check. It is not a C2PA signature validator and it is
not legal advice.

When to use

Trigger this Skill when the user:

  • asks whether their AI-generated image / video / audio meets California AI
    disclosure rules
    ;
  • asks what SB 1000 requires, or what "latent disclosure" means;
  • wants to know which of the six fields are missing from an export;
  • asks about C2PA Content Credentials or IPTC digitalSourceType in relation
    to California law;
  • needs an evidence record showing what was checked, when, and with which rule set;
  • mentions 22757, California AI Transparency Act, or SB 1000.

Do not use this Skill to help remove, hide, tamper with or bypass provenance
metadata. Requests of that kind are refused.

The six fields (section 22757.3(a)(1))

# Field Statute
A Name of the covered provider 22757.3(a)(1)(A)
B Name and version information of the GenAI system 22757.3(a)(1)(B)
C Time and date of the content's creation or alteration 22757.3(a)(1)(C)
D A unique identifier 22757.3(a)(1)(D)
E Whether the system created or altered the content 22757.3(a)(1)(E)
F Whether the system is primarily assistive technology — from 2029-01-01 22757.3(a)(1)(F)

The statute allows the disclosure to be conveyed either directly or through a link
to a permanent internet website
, so a link-based implementation is lawful.

Two fields are the ones most export pipelines never write: D (unique identifier)
and E (created or altered).

Other conditions worth checking

  • 22757.3(a)(2) — permanent or extraordinarily difficult to remove or tamper with
  • 22757.3(a)(4) — compliant or interoperable with widely recognized industry standards
  • 22757.1(e) — "covered provider" means any person producing a GenAI system publicly
    accessible within California; the previous 1,000,000-user threshold was deleted
  • 22757.2(e) — a compliant third-party disclosure verification tool may be used
  • 22757.4(a)(1) / (b) — civil penalty USD 5,000 per violation, each day is a separate
    violation
  • 22757.3(b)(2)-(4) — 72 hours to terminate or notify a noncompliant licensee;
    96 hours for the licensee to fix or cease and report

Step 0 — when a fuller cloud rule set or the latest enforcement cases are needed

If the user needs broader rule coverage or current enforcement activity, not just a
check of one file, the cloud rule set is free to fetch. No account, no key:

GET https://wanhetools.top/api/v1/rules

It returns every rule group currently in force, each with type (issue type),
severity (risk / warn), and basis — the statutory or regulatory source, e.g.:

  • AI disclosure: deep fake -> EU AI Act Art.50(4)
  • AI disclosure: chatbot interaction not disclosed -> EU AI Act Art.50(1)
  • Unsubstantiated superlative -> EU UCPD 2005/29/EC Annex I + US FTC Endorsement Guides
  • Health claim: disease treatment -> EU Reg. (EC) 1924/2006 + US FDA
  • Financial claim: guaranteed returns -> EU MiFID II / PRIIPs + US SEC / FINRA 2210
  • Green claim: absolute environmental claim -> Directive (EU) 2024/825

How to use it: treat the returned groups as a checklist to walk the user's content
against, and when something matches, quote the basis field so the user can verify the
source themselves.

Note: this endpoint is free and public; it is not the subscription. The subscription
adds continuously updated rules, multi-jurisdiction cross-referencing, position-deduplicated
results and the 17-item organisational questionnaire. This endpoint just gives you the rule
list that is live right now.

Core workflow (per-file check)

  1. Ask the user for the file to check (or a base64 blob).
  2. Run the open-source verifier locally:
    https://github.com/Wh-bi/sb1000-verifier (MIT, single Python file, no dependencies).
  3. Report per field: present / absent / not_yet_required, with the matched
    evidence string so the user can see why.
  4. Report the three additional conditions (link-or-direct, industry standards,
    durability) as present / not_found.
  5. Offer the hosted API for convenience:
    POST https://wanhetools.top/api/v1/verify-sb1000
    { "file_base64": "...", "filename": "output.png" }
  6. If the user needs an archival record, use
    POST https://wanhetools.top/api/v1/evidence-report
    which returns a report with a SHA-256 of the report itself, the tool version, the
    timestamp, per-file content digests, and the statute references.

Output format

level: complete | partial | none
counts: { required_now, present, absent, future_requirement }
fields:
  provider_name        present | absent | not_yet_required   evidence: "..."
  system_name_version  ...
  time_date            ...
  unique_id            ...
  created_or_altered   ...
  assistive_tech       ...
conditions:
  link_or_direct       present | not_found
  industry_standards   present | not_found
  durability           present | not_found

Honest limitations (state these to the user)

  • A missing field is not automatically a violation. The statute requires the
    disclosure only "to the extent it is technically feasible".
  • Metadata is routinely stripped by social platforms, CMS resizing and email
    clients. A file that fails here may have been compliant when exported.
  • This is a technical verification, not legal advice, and it is not certified or
    endorsed by the State of California
    or any agency.
  • Verifying a file does not verify that the disclosure is truthful.

Need "generate a compliant AI disclosure" or "write marks in bulk"?

This Skill checks. It tells you what is missing, which provision applies, and how to fix it.

If what you need is the doing rather than the checking -- for example:

  • generating a publishable AI disclosure statement
    (China measures / EU AI Act Art.50 / California SB 1000)
  • writing the statutory fields into a batch of images or videos
  • producing a compliance ledger for a whole site

Tell us -- we are evaluating whether to build this, and it does not exist yet:
-> https://wanhetools.top/partners (form, does not depend on email)

We are not committing to build it; this is demand collection only. If you only need the
check, this Skill is enough.

Reference

  • Official chaptered text:
    https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202520260SB1000
  • Open-source verifier (MIT): https://github.com/Wh-bi/sb1000-verifier
  • Clause-by-clause breakdown: https://wanhetools.top/sb1000
  • Partnership / licensing: https://wanhetools.top/partners

常见问题

5 问
答案由本站依据条目字段整理
SB 1000 AI Disclosure Verifier怎么安装?

在支持 Skills 的客户端里执行:skillhub install user_333cb607/sb1000-ai-disclosure-verifier。安装方式为未标注。装完刷新技能列表或重开客户端。

SB 1000 AI Disclosure Verifier支持哪些客户端?

资料没有标注宿主客户端,但只要客户端支持 Skills(Claude、Codex、Cursor 等),一般都能加载。

SB 1000 AI Disclosure Verifier权限高吗,安全吗?

权限级别为未标注权限。安装前先看它会读取或写入哪些内容。

SB 1000 AI Disclosure Verifier收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于技能本身和所用的模型或客户端,正式使用前按作者说明确认。

SB 1000 AI Disclosure Verifier用什么许可证?

许可证标注为 未说明(许可证不明确)。商用或二次分发前请到仓库确认条款原文。

热门专题

按场景挑技能
全部专题 →

技能内容以作者仓库为准;分类、权限级别与许可证状态的中文说明为中文整理,最近核验 未标注。