Acceptable Use Policy

1. Scope

This Policy applies to all content processed by the Service, including:

• Text: AI-generated translations, output documents, glossaries and API responses.
• User input: documents, prompts, instructions and terminology you submit.
• Account activity: API usage patterns and interactions with the platform.

2. Prohibited Content Categories

You must not use the Service to translate, generate, store or distribute any of the following six categories of content:

2.1 Absolute prohibitions — zero tolerance

The following are prohibited without exception. We act immediately and reserve the right to report to authorities:

• Child sexual abuse material (CSAM) or any content that sexualises a minor, including uploading photographs or documents depicting minors for sexualised translation.
• Content promoting or assisting terrorism, extremism or mass violence.
• Technical instructions for manufacturing weapons of mass destruction.
• Content inciting genocide or ethnic hatred.
• Non-consensual sexual deepfakes or impersonation of a real person for fraudulent or abusive purposes.
• [ADD ANY CATEGORY REQUIRED BY YOUR TARGET MARKET'S REGULATION]

There is no exception to Section 2.1. Content that appears to sexualise a minor is refused, the account is immediately suspended and permanently terminated, we preserve all available evidence, and we report suspected CSAM to the relevant law enforcement authorities and child protection bodies (including NCMEC) without delay and irrespective of where the account is located. We cooperate fully with lawful requests.

2.2 Restricted — permitted only under stated conditions

• Adult content: not permitted. DocTranslator is not an adult content platform and has no age verification pathway for such material.
• Violent content: permitted only where an explicit documentary, journalistic or educational context applies, and in all cases the Section 2.1 zero-tolerance rules still apply.
• Medical, legal and financial content: permitted for translation purposes only. Output is reference material and must not be presented as professional advice (see Section 9.4 of the Terms).
• Politically sensitive content: permitted for translation, provided it is not misleadingly synthetic or designed to deceive.

2.3 General prohibitions

• Translating content you do not have the right to translate, or that infringes a third party's copyright, trademark or other rights.
• Uploading malware, viruses, malicious code, or using the Service for phishing, network attacks or security testing without written authorisation.
• Spam, harassment, or bulk abusive activity.
• Reselling, sublicensing or redistributing access, or sharing a single account across unrelated parties.
• Scraping, rate-limit circumvention, or automated bulk collection of output beyond your authorised use.
• Circumventing or interfering with our security controls, moderation systems, quotas or rate limits.
• Using jailbreak or obfuscation techniques to evade this Policy.
• Using the Service's inputs, outputs or derivative data to train, fine-tune, benchmark or distil any AI model that competes with [LEGAL COMPANY NAME]'s products.
• Presenting AI-generated output as human-authored in a way that misleads, or as the work of a licensed professional where that could cause harm.

3. Content Moderation

We operate a layered "automated screening + human review + periodic audit" moderation system. We never pass raw, unmoderated model output directly to users.

3.1 Automated screening (real time)

• Input filtering: uploaded documents and prompts are scanned for prohibited categories, including a maintained blocklist covering CSAM terminology, sexual content involving minors, weapons, and known infringing material references.
• Output filtering: generated translations are screened before being returned to you.
• Behavioural analysis: anomalous API usage patterns and repeated policy hits are flagged for review.
• Moderation tooling: [NAME YOUR REAL MODERATION TOOLING, e.g. keyword + semantic classifier with DeepSeek-based content screening; Waffo Prompt Sift where enabled]. We actively update the blocklist — a static, one-off word list is not our control.
• Audit logs: automated moderation decisions and human review outcomes are retained as moderation records and are available for review on request.

3.2 Human review

• Content flagged automatically enters a human review queue.
• User reports are reviewed by our trust and safety team.
• Reviewers complete content safety training and are bound by confidentiality obligations.

3.3 Periodic audit

• We perform random sampling reviews of moderation decisions on a recurring basis.
• We evaluate and improve automated screening accuracy based on review outcomes.

3.4 Severity classification

4. Reporting Violations

4.1 When to report. Report content or behaviour that breaches this Policy, content involving child safety or terrorism, or AI impersonation of you or your organisation.

4.2 Our reporting principles.

• Confidentiality: we treat your identity as a reporter confidentially.
• Impartiality: every report is assessed independently and objectively.
• Anti-abuse: malicious or knowingly false reports are recorded and may themselves lead to enforcement action.

5. Reporting Channels

Please use the channel that best fits your report. Only list channels you genuinely monitor.

• Safety and urgent reports: report@pandada.ai
• Content safety team: safety@pandada.ai
• Appeals: appeal@pandada.ai
• General support: support@pandada.ai
• Online form: [ADD AN ONLINE REPORT FORM URL IF AVAILABLE]

When reporting, please include the account email, the document or output concerned, and a description of the policy section you believe was breached. For L1 child safety reports, mark the subject line "URGENT - CHILD SAFETY".

6. Response and Resolution Timelines

If resolution requires longer, we will proactively notify the reporter with a reason and a revised timeframe. Process: report received (auto-acknowledged) → severity classification (L1–L4) → review (automated + human) → decision → enforcement executed → outcome communicated.

7. Enforcement Actions

We apply proportionate actions from the following ladder. We only take measures we are technically able to execute.

Where we terminate an account, we will, where appropriate, refund prepaid unused amounts. We may withhold details of our investigation to protect other users and our systems.

8. Appeals

If you disagree with an enforcement decision, you may submit an appeal within 14 days of notification:

• Channel: appeal@pandada.ai

Appeals are reviewed within 10 business days by a reviewer who was not involved in the original decision. We will confirm the outcome in writing.

9. Policy Changes

We may update this Policy. Material changes will be notified at least 15 days in advance by platform notice or email. Continued use after the effective date constitutes acceptance. We will publish a version history on request so the policy in force at any past date can be evidenced.

10. Contact Us

• Content safety: safety@pandada.ai
• Reports: report@pandada.ai
• Appeals: appeal@pandada.ai
• Legal name: [LEGAL COMPANY NAME]
• Registered address: [REGISTERED ADDRESS]