Skip to main content

Acceptable Use Policy

GraffForge acceptable use policy: prohibited AI content, prompt screening and image moderation before display, fail-closed enforcement, and abuse reporting.

Scope

This Acceptable Use Policy applies to prompts, requested text, reference images, AI-generated images, and content stored in GraffForge History. It complements the Terms of Service and License & Usage Rights.

GraffForge is an independent product. AI images are generated through the APIMart (api.apimart.ai) model gateway using gemini-nano-banana-2.1. GraffForge is not an official APIMart or model-provider product, and those providers do not endorse GraffForge.

Prohibited generation categories

GraffForge does not allow generation of:

  • Pornography or NSFW content.
  • Violence or gore.
  • Hate speech, hateful imagery, or content inciting genocide or ethnic hatred.
  • Child-unsafe content, including child sexual abuse material (CSAM) or any sexual content involving minors.
  • Deepfakes or impersonation of real people, including non-consensual sexual deepfakes.
  • Content that infringes copyright or trademarks, including deceptive use of a protected character, logo, brand, or artwork.

We also prohibit content that promotes or facilitates terrorism, extremism, mass violence, or weapons of mass destruction; deceptive spam or fraud; and attempts to bypass GraffForge safety controls.

Moderation

GraffForge uses two enforced checks for production AI generation:

  1. Before generation, user-supplied text is screened by Waffo Prompt Sift with semantic enforcement. Only an allow verdict proceeds to the image model.
  2. After generation and before storage or display, the generated image is classified by Cloudflare Workers AI using Meta Llama 3.2 11B Vision Instruct. Only an allow verdict is saved and displayed.

If either service returns block, review, an unreadable result, or is unavailable, GraffForge fails closed: the image is not displayed and the reserved generation is returned. GraffForge stores redacted moderation records—stage, verdict, reason, matched category identifiers, and correlation ID—but not a second copy of prompt text or image bytes in the moderation log.

GraffForge has no public gallery. Saved results are private to the account that created them. User reports are reviewed manually by the GraffForge operator.

Automated moderation can make mistakes and does not replace your obligation to submit lawful content or review output before use.

Reporting

Report suspected violations, child-safety concerns, impersonation, or infringement to support@graffforge.com. Include the relevant GraffForge job ID or page, a short explanation, and a way to contact you. Do not email illegal files; describe the content and where it can be found.

The support inbox is monitored by the GraffForge operator. Critical child safety, terrorism, or credible mass-violence reports are prioritized. Other reports are handled according to severity and available evidence. GraffForge does not claim a 24/7 moderation team.

Enforcement

Depending on severity and recurrence, GraffForge can block a request, remove a saved generation, issue a warning, revoke unused generation allowance, suspend an account, or permanently ban an account. Apparent criminal activity may be preserved or reported when required by law. Zero-tolerance violations may lead to immediate removal and a permanent ban.

Appeals

If you believe a request or account was actioned incorrectly, email support@graffforge.com within 14 days. Include the moderation request ID or job ID if available and explain why the decision should be reconsidered. Appeals are reviewed manually; we will correct the record or restore access when the evidence supports it.

Changes and contact

Material changes, including a change to the underlying production image model or moderation tools, will be reflected on this page. Safety and policy questions can be sent to support@graffforge.com.