制定服务专属可接受使用政策
结合服务功能与团队执行能力识别滥用风险,起草使用规则、举报申诉流程和分级处置方案。
完整提示词
<context>
An acceptable use policy (AUP) is the rulebook the terms of service point to. It works when its prohibitions match what the service can actually be abused for: a file host worries about malware, illegal content and copyright; an email tool about spam and phishing; a compute platform about crypto-mining, scanning and attacks; a community about harassment and doxxing. Generic AUPs copied from a hyperscaler list fifty things nobody enforces and miss the abuse this product attracts. Two other common faults: the policy promises enforcement steps the team cannot deliver, and it has no route for users to report abuse or appeal a decision. In several places, rules for online platforms now set duties on notice handling, statements of reasons and appeals (for example the EU Digital Services Act for intermediary services); which apply depends on the service type and size, so treat them as points to confirm.
Jurisdictions: {{jurisdictions}}
</context>
<task>
Service:
<service>
{{service}}
</service>
{{#enforcement_capacity}}
Enforcement today:
<enforcement>
{{enforcement_capacity}}
</enforcement>
{{/enforcement_capacity}}
1. Risk map: from the features described, list the realistic abuse types (each tied to a feature), how likely and how harmful each is, and whether it is illegal content, a security abuse, harm to others, or a resource or commercial abuse.
2. Draft the policy in plain language:
- Purpose and who it applies to, including users' own end users or team members where the service is resold or shared.
- Prohibited uses grouped by the risk map: illegal content and activity, harm to others (harassment, threats, doxxing, impersonation), security abuse (malware, scanning, credential attacks, circumventing limits), spam and deceptive practices, infringement, and resource abuse. Give a concrete example for each, specific to this service.
- Rules for automated access and the API, if any.
- How to report abuse: a channel in [BRACKETS], what to include, and a separate urgent route for content involving children or imminent harm.
- What the provider may do: remove content, restrict features, suspend, terminate, preserve records, and report to authorities where required.
- Notice, reasons and appeals: what the user is told and how to contest a decision.
- Changes to the policy and how users are notified.
3. Enforcement ladder: a table matching violation severity to the first response, with an immediate-action tier for illegal or dangerous content, sized to the team's real capacity.
4. Points to confirm: every legal duty assumed (notice and action, statements of reasons, reporting obligations, data retention for investigations) marked "confirm for [jurisdiction]".
</task>
<constraints>
{{> guardrails/professional-limits}}
- Tie every prohibition to the service's features; do not pad the list with abuse that cannot happen on this service.
- Do not promise response times or reviews the stated team cannot deliver; if capacity is unknown, use [TIMEFRAME] placeholders.
- Do not cite statutes, article numbers or thresholds as fact unless the user supplied them; describe the obligation and mark it to confirm.
- Never invent contact addresses, regulator names or law enforcement channels; use [BRACKETS].
- Keep reporting to authorities factual: say the provider may report where the law requires or permits, without overstating it.
{{> output/uncertainty}}
</constraints>
<output_format>
## Risk map
Table: feature | abuse type | likelihood | harm | category.
## Policy
The full AUP with headings.
## Enforcement ladder
Table: severity | examples | first response | repeat response | who decides.
## Points to confirm
Numbered.
</output_format>可替换内容
- {{jurisdictions}}
- {{service}}
- {{#enforcement_capacity}}
- {{enforcement_capacity}}
- {{/enforcement_capacity}}
- {{> guardrails/professional-limits}}
- {{> output/uncertainty}}
- [BRACKETS]
- [jurisdiction]
- [TIMEFRAME]
分类与标签
办公效率、平台规则制定、服务合规、滥用处置