1. Scope
This Acceptable Use Policy ("Policy") applies to all content generated, published, disseminated, or stored through PDF.ai's services, including but not limited to PDF parsing, field extraction, document Q&A, and all AI-powered features. All users, developers, and merchants using our platform must comply with this Policy.
This Policy covers the following content forms:
- Text: AI-generated text, dialogue, code, articles, etc.
- Images: AI-generated or edited images, illustrations, etc. (if applicable)
- Audio/Video: AI-synthesized voice, music, video (if applicable)
- User Input (Prompt): Prompts and instructions submitted by users
2. Content Standards and Prohibited Categories
PDF.ai explicitly prohibits the use of our services to generate the following six categories of content:
- Pornography / NSFW content
- Violence / gore
- Hate speech
- Child Sexual Abuse Material (CSAM)
- Deepfakes / impersonation of others
- Content infringing copyright or trademarks
2.1 Absolute Prohibitions — Zero Tolerance
The following content will result in immediate action upon discovery, and we reserve the right to report to authorities:
- Child sexual exploitation material (CSAM) or any sexual or exploitative content involving minors (persons under 18)
- Content promoting or assisting terrorism, extremism, or mass violence
- Instructions for manufacturing weapons of mass destruction
- Content inciting ethnic genocide or national hatred
- Non-consensual deepfake sexual content targeting real individuals
- [Additional categories per your target market regulations]
2.2 Strictly Restricted — Requires Review
- Adult content: Only on compliant platforms with age verification completed
- Violent content: Must have clear creative or educational context
- Politically sensitive content: Potentially misleading content
- Medical/Legal/Financial advice: Must comply with local regulations
2.3 General Prohibited Activities
- Generating or spreading false information, rumors
- Impersonating real individuals, brands, or institutions
- Infringing intellectual property, portrait rights, or privacy rights
- Deceiving users by presenting AI content as human-created
- Batch generating spam or fraudulent content
- Using jailbreak prompts to bypass platform security mechanisms
3. Content Moderation Mechanism
PDF.ai employs a multi-layer moderation system: Automated Moderation + Human Review + Regular Sampling.
3.1 Automated Moderation (Real-time)
All requests and generated content are scanned in real-time by security filters:
- Input filtering: Keyword, semantic, and intent recognition on prompts
- Output filtering: Real-time security assessment of generated content
- Usage pattern analysis: Monitoring for abnormal API call behavior
- Security models used: [Specify, e.g., OpenAI Moderation API or proprietary model]
3.2 Human Review
- Content flagged by automation enters the human review queue
- User-reported content is manually verified by our Trust & Safety team
- Review team: [e.g., 7×24 on-call review team]
- Reviewers must complete content safety training and sign confidentiality agreements
3.3 Regular Sampling and Audits
- Random sampling review conducted every [frequency, e.g., monthly]
- Regular evaluation and optimization of automated moderation system accuracy
- [Whether third-party independent audits are conducted, if applicable]
3.4 Content Classification Standards
| Level |
Description |
Review |
Action |
| L1 Critical |
Child safety, terrorism |
Auto-block + Human confirmation |
Immediate ban + Report |
| L2 High Risk |
Serious violations |
Auto-flag + Priority review |
Remove + Suspend account |
| L3 Medium Risk |
General violations |
Human review |
Warning + Remove |
| L4 Low Risk |
Minor violations |
Report-triggered |
Warning + Request modification |
4. Reporting Mechanism
4.1 Reportable Situations
- Content violating this Policy's content standards
- Content involving child safety, terrorism, or other serious violations
- Discovery of AI impersonation or infringement of one's own rights
- [Other situations the platform deems reportable]
4.2 Reporting Principles
- Confidentiality: Reporter identity is strictly protected
- Neutrality: All reports undergo independent, objective review
- Anti-abuse: Malicious or false reports will be recorded and handled
5. Reporting Channels
At least one real, monitored channel must be retained:
6. Response Timeframes
| Violation Type |
Initial Response |
Resolution Complete |
| L1 Critical | Within 2 hours | Within 24 hours |
| L2 High Risk | Within 24 hours | Within 3 business days |
| L3 Medium Risk | Within 3 business days | Within 7 business days |
| L4 Low Risk | Within 5 business days | Within 15 business days |
If extended processing time is needed, we will proactively notify the reporter with an explanation. Processing flow: Report received (auto-confirm) → Initial classification (L1–L4) → Content review (auto + human) → Action decision → Execute action → Notify result.
7. Enforcement Actions
| Action |
Measure |
Applicable Scenario |
| 1. Content Warning | Written warning, request deletion or modification | First minor violation |
| 2. Content Removal | Remove violating content | Confirmed L3–L4 |
| 3. Feature Restriction | Reduce call limits or model access | Multiple minor or single L3 |
| 4. Account Suspension | Temporarily suspend pending review | L2 or accumulated violations |
| 5. Permanent Ban | Permanently terminate access | L1 or multiple L2 |
| 6. Legal Action | Report to law enforcement, cooperate with investigation | L1 involving illegality |
8. Appeals Mechanism
If you disagree with an enforcement result, you may submit an appeal within 30 days of receiving notification:
- Appeal channel: appeal@pdf.ai
- Processing timeframe: Within 10 business days for review completion
- Review method: Re-examined by an independent reviewer not involved in the original decision
9. Transparency and Policy Updates
- We publish a content safety transparency report every [frequency, e.g., quarterly]
- Material changes will be notified at least 15 days in advance
- Continued use of the platform constitutes acceptance of the updated Policy