← 返回 Skills 市场

AI Governance & Safety Intelligence

Name: AI Governance & Safety Intelligence
Author: ai-gaoqian

作者 ai-gaoqian · GitHub ↗ · v1.0.0 · MIT-0

cross-platform ✓ 安全检测通过

总下载

当前安装

版本数

在 OpenClaw 中安装

/install ai-governance-intel

功能描述

Provides comprehensive analysis and comparison of global AI regulations, safety incidents, model evaluations, standards, and international governance for pol...

使用说明 (SKILL.md)

AI Governance & Safety Intelligence

The definitive AI governance intelligence engine. 25 authoritative sources covering every major regulatory framework (EU AI Act, US EO 14110, China GenAI Measures, OECD AI Observatory), 3,000+ AI incident records, frontier model evaluation benchmarks, and international cooperation frameworks. Built for policymakers, compliance officers, AI safety researchers, and anyone tracking the rapidly evolving AI governance landscape.

Triggers

"What AI regulations apply to [use case]?"
"Track EU AI Act compliance milestones"
"Compare AI governance: US vs. EU vs. China"
"Analyze latest AI safety incidents"
"Assess frontier model evaluation frameworks"
"Monitor international AI safety summit outcomes"
"Check ISO 42001 certification requirements"
"Research AI liability frameworks"

Capabilities

#	Capability	Data Scope	Output Mode
1	Regulatory jurisdiction comparison	45+ jurisdictions: EU AI Act, US EO 14110, China GenAI Measures, UK AISI, CAIDP rankings	Side-by-side table: risk classification, obligations, enforcement, penalties
2	Compliance timeline tracking	EU AI Act 2024-2027 phased rollout, global milestones	Gantt-style timeline: prohibited practices → GPAI rules → high-risk obligations → full enforcement
3	AI incident database analysis	AIID (3,000+ incidents), OECD AIM, MIT 777-risk taxonomy	Incident trending by harm type, root cause, deployer, sector, severity
4	Frontier model safety evaluation	OpenAI Preparedness, Anthropic RSP (ASL 1-4), DeepMind Frontier Safety, METR, ARC Evals	Capability threshold comparison, CBRN/cyber/autonomy benchmarks, evaluation methodology gap analysis
5	Standards & certification landscape	ISO/IEC 42001, ISO 23894, IEEE 7000, CEN-CENELEC JTC 21, BSI Hub	Certification roadmap, conformity assessment bodies, standard-to-regulation mapping
6	International AI governance cooperation	UN AI Advisory Body, AI Safety Summit series (2023-2026), G7 Hiroshima AI Process, Global AISI Network (10+ institutes)	Institutional mapping, mandate comparison, soft law vs. hard law trajectory
7	AI safety & alignment research landscape	Anthropic, DeepMind, OpenAI, ARC Evals, METR, Epoch AI, CAIS, Redwood	Research program comparison, key papers, open problems, scaling trajectory data
8	Compute governance & export controls	BIS Entity List, GPU export restrictions, chip tracking proposals	Hardware restriction tiers, impact assessment on AI capability diffusion, sovereign AI compute gap
9	Digital rights & AI ethics	Access Now, AlgorithmWatch, STOP	Facial recognition bans, predictive policing, automated decision audits, civil society position mapping
10	AI risk horizon scanning	Integrated cross-source fusion	Emerging risks: agentic AI, open-weight models, bio risk, labor displacement, elections integrity

Workflow

User Query
    │
    ├─ Regulatory compliance query → Jurisdiction comparison (Capability #1) + Timeline (Capability #2)
    │   └─ Output: Applicable frameworks table + compliance milestones + penalty exposure
    │
    ├─ Safety evaluation query → Frontier model eval (Capability #4) + Incidents (Capability #3)
    │   └─ Output: RSP/Preparedness level mapping + relevant incident patterns + residual risk assessment
    │
    ├─ International governance query → Cooperation tracking (Capability #6) + Standards (Capability #5)
    │   └─ Output: Institutional landscape + mandate mapping + harmonization gap analysis
    │
    ├─ Open-source AI debate query → Compute governance (Capability #8) + Digital rights (Capability #9)
    │   └─ Output: Pro/con analysis with evidence, regulatory trajectory in key jurisdictions
    │
    └─ Horizon scanning → Risk scanning (Capability #10) + Research landscape (Capability #7)
        └─ Output: Emerging risk dashboard + research direction watchlist + 12-month outlook

Output Formats

Format 1: Regulatory Comparison Matrix

| Jurisdiction | Risk Classification | Conformity Assessment | Transparency | Enforcement | Penalties |
|-------------|--------------------|----------------------|--------------|-------------|-----------|
| EU AI Act   | Unacceptable/High/Limited/Minimal | Third-party (High) | ✓ | EU Commission + MS | Up to 7% global turnover |
| US EO 14110 | Federal agency RMF | Self-assessment (voluntary) | ✓ (exec. branch) | NIST + OMB | Agency-level |
| China GenAI | Filing + algorithm registry | CAC review | ✓ | CAC | Service suspension |

Format 2: Frontier Model Safety Dashboard

| Model | Developer | Framework | CBRN Risk | Cyber Risk | Autonomy Risk | Persuasion Risk |
|-------|-----------|-----------|------------|------------|---------------|-----------------|

Format 3: AI Incident Trend Analysis

| Harm Type | 2023 | 2024 | 2025 | Trend | Most Affected Sector |
|-----------|------|------|------|-------|---------------------|

Usage Guidelines

Always cite regulatory source with effective date
Distinguish between enacted law, proposed regulation, and voluntary framework
Flag regulatory divergence as risks for cross-border deployment
Use AI incident data responsibly; note that underreporting biases exist
When assessing frontier models, reference the developer's own framework

Examples

Q: Do I need to comply with EU AI Act for my chatbot? → EU AI Act Article 5 (prohibited) + Article 6 (high-risk) + Article 52 (transparency for chatbots) → Classification: [transparency obligation only], requirements: [user disclosure, no other obligations], timeline: [effective now, full 2026]

Q: Compare US vs. China AI chip restrictions → BIS Entity List (SMIC, Huawei) + semiconductor export tiers + CAC algorithm registry → US restricts [GPU tiers T1-T3], China restricts [training data, algorithm filing], impact: [quantified]

Q: Latest frontier model safety incidents → AIID database query + OECD AIM + developer transparency reports → Top 3 incident categories: [deepfakes elections / autonomous agent failures / bias discrimination], trend: [increasing/decreasing]

Data Sources (25 total): See references/ai_governance_sources.json for complete listing with URLs and update frequencies. （内容由AI生成，仅供参考）

安全使用建议

Installers should treat this as an informational research aid, not legal advice. Because AI regulation changes quickly, verify important compliance conclusions against primary official sources and current effective dates.

能力评估

✓ Purpose & Capability

The stated purpose is AI governance, regulation, safety incident, and standards analysis; the artifacts contain a SKILL.md with matching workflows and a JSON list of public reference sources.

✓ Instruction Scope

Runtime instructions are limited to research-style outputs, citations, comparisons, and caveats such as distinguishing enacted law from proposals; there are no instructions to override user control, hide behavior, or perform unrelated actions.

✓ Install Mechanism

The package contains only markdown and JSON reference data, with no declared dependencies, executable scripts, API-key requirement, or install-time commands.

ℹ Credentials

The skill points agents toward many external public sources and current regulatory topics, which is expected for its research purpose; users should still verify current law and dates before relying on outputs.

✓ Persistence & Privilege

No persistence, background workers, credential use, local file indexing, privilege escalation, or mutation authority is present in the artifacts.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install ai-governance-intel
安装完成后，直接呼叫该 Skill 的名称或使用 /ai-governance-intel 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

Initial release: Comprehensive AI governance intelligence engine. - Covers 25 authoritative sources, 3,000+ AI incident records, and major regulatory frameworks (EU AI Act, US EO 14110, China GenAI Measures, OECD, UK, etc.). - Supports regulatory comparison, compliance timeline tracking, AI safety/incident analysis, frontier model evaluation, and international cooperation monitoring. - Provides structured output formats (comparison matrices, incident trends, safety dashboards). - Designed for policymakers, compliance officers, researchers, and anyone tracking AI governance and safety. - Includes clear usage guidelines and practical examples for regulations, incidents, and risk assessments.

元数据

Slug ai-governance-intel

版本 1.0.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题