Markdown Editor
Mở trình soạn thảo

Bộ mẫu nhà cung cấp mô hình GPAI

Nghĩa vụ của nhà cung cấp AI đa dụng: tài liệu mô hình, chính sách bản quyền, tóm tắt dữ liệu huấn luyện, rủi ro hệ thống và minh bạch.

9 mẫu trong 3 bước, theo thứ tự sử dụng. Mở mẫu bất kỳ trong trình soạn thảo, điền nội dung rồi xuất ra PDF, DOCX hoặc HTML.

Bước 1

Tài liệu

  1. GPAI Model Technical Documentation Template (EU AI Act Annex XI and XII)

    Đầy đủ

    Document a general-purpose AI model per EU AI Act Article 53, Annexes XI and XII, copyright policy and systemic risk

    Regulation (EU) 2024/1689 as amended by Regulation (EU) 2026/1744, Article 53 and Annexes XI and XII

  2. GPAI Copyright Policy Template

    Chuẩn

    EU AI Act Art. 53(1)(c) copyright policy for GPAI providers: lawful crawling, robots.txt and TDM opt-outs, crawler transparency, output safeguards, complaints

    EU AI Act Art. 53(1)(c); GPAI Code of Practice Copyright chapter (Jul 2025)

  3. GPAI Public Summary of Training Content Template

    Chuẩn

    EU AI Act Art. 53(1)(d) summary: provider and model, modality sizes, data sources by category, domain list, TDM opt-outs, illegal content

    EU AI Act Art. 53(1)(d); Commission training content template (July 2025)

  4. Model Card Template (Mitchell et al. 2019)

    Đầy đủ

    Model details, intended and out-of-scope uses, factors, metrics, data, disaggregated results, ethics, caveats

    Model Cards for Model Reporting (Mitchell et al., 2019)
Bước 2

Đánh giá tác động

  1. Frontier AI Safety and Security Framework Template (GPAI Code, SB 53)

    Đầy đủ

    Applicability tests, systemic risk register, evaluations, capability tiers, safety and weight security, deployment decisions, incident clocks, whistleblowing

    GPAI Code of Practice Safety and Security chapter (Jul 2025); EU AI Act Art. 51 to 55; California SB 53; New York RAISE Act (amended 2026)

  2. AI Red Team Plan and Report Template (OWASP LLM Top 10 2026, Agentic ASI)

    Đầy đủ

    Red team scope, threat model, NIST AI 100-2 classes, OWASP LLM 2026 and ASI01-ASI10 coverage, MITRE ATLAS, test cases, severity, findings, retest

    OWASP Top 10 for LLM Applications 2026; OWASP Top 10 for Agentic Applications 2026; MITRE ATLAS (monthly, v2026.10); NIST AI 100-2 E2025

  3. AI Incident Report Template

    Chuẩn

    AI incident or near miss: classification, Art. 3(49) serious incident test, timeline, harm, evidence, root cause, Art. 73 deadlines, notifications, CAPA

    OECD AI incident concepts (2024); EU AI Act Articles 3(49), 26(5), 73
Bước 3

Công bố và thông báo bảo mật

  1. AI Content Transparency and Marking Plan Template (AI Act Art. 50, C2PA)

    Đầy đủ

    Article 50 role and scope, interaction disclosure, C2PA and watermark marking, robustness tests, labels, exceptions, Code of Practice mapping, China labelling

    EU AI Act Art. 50 (Reg. (EU) 2024/1689 as amended by 2026/1744); Code of Practice on Transparency of AI-generated Content (2026); C2PA 2.5

  2. Hiroshima AI Process Transparency Report Template (OECD HAIP v2.0)

    Đầy đủ

    OECD HAIP Reporting Framework v2.0: role, risk identification, risk management and security, transparency, governance, provenance, research, societal benefit, evidence

    OECD HAIP Reporting Framework v2.0; G7 Hiroshima Code of Conduct (2023)