凡自治之物 · LGD 治理学 · 创造术家族 ｜ 评测协议：见平台质量门禁规范
English
When to use
- Before a government AI decision-support system generates policy advice, to verify the data source holds a legal authorization basis.
- When a public-sector auto-reply bot touches sensitive information, to trigger the three-laws gate determination.
- Before internal data-governance tools process cross-department sharing requests or desensitized output, to run autonomous identity and permission audits.
- When an AI app boots in self-check mode to confirm its own logic does not rely on external permissions or fall short of the LGD standard.
Usage
- Run
python gov_guard.pydirectly as the main entry point — no extra environment dependencies. - Core parameters:
--input_data(the text or dataset under evaluation) and--mode(decision / reply / governance). - To read the logic, inspect the three-laws determination functions and rule set defined in
gov_guard.py.
Workflow
- Load the government AI content or data stream to be checked into a temporary workspace.
- The script initializes the LGD three-laws (registered · evidenced · gated) rule engine and sets thresholds.
- The system independently evaluates data source, authorization evidence and access permissions in parallel.
- Produce the internal compliance score and the verdict on whether the zero-dependency standard is met.
- Emit an audit report with detailed risk points to the standard log file.
- Based on the verdict, automatically release or block the downstream AI task.
Evidence & safety statement
- Evidence: TBD (no evaluation / badge link yet; update here once it is produced)
- Version: 1.0.1
- Security: no independent security scan has been performed (TBD); the conclusion is pending and no endorsement is given until it is completed.
中文
何时用
- 政务 AI 决策辅助系统在生成政策建议前，需校验数据来源是否具备法定授权依据。
- 公共部门面向公众的自动回复机器人涉及敏感信息时，需触发三律门禁判定。
- 内部数据治理工具在处理跨部门共享请求或隐私脱敏输出前，进行自主身份与权限审计。
- AI 应用启动自检模式以确认其自身逻辑未违规依赖外部权限及不符合 LGD 标准。
用法
- 直接调用
python gov_guard.py作为主入口脚本运行，无需额外环境依赖。 - 核心参数包括：
--input_data(待评估的文本或数据集),--mode(决策/答复/治理)。 - 若需读取逻辑细节，可直接阅读分析
gov_guard.py源码中定义的三律判定函数与规则集。
流程
- 输入端加载待校验的政务 AI 内容或数据流至临时工作区。
- 脚本初始化 LGD 三律（有籍·有证·有门禁）规则引擎并配置阈值。
- 系统依次对数据来源、授权证明、访问权限进行独立并行评估。
- 生成内部合规评分及是否符合“零依赖”标准的判定结果。
- 输出包含风险点明细的审计报告至标准日志文件。
- 根据判定结果自动触发放行或阻断后续 AI 任务执行动作。
证据与安全声明
- 证据：TBD（暂无评测/徽章链接；产生后在此更新实际链接）
- 版本：1.0.1
- 安全：未做独立安全扫描（TBD），结论待补；补全前不作推荐背书。