Create responsibly.
Updated October 28, 2026. These rules cover prompts, uploaded images and generated videos, including any audio in a generated video.
Content we do not allow
- Pornography, sexually explicit or NSFW content, including adult sexual content and non-consensual intimate imagery.
- Violence or gore, including graphic injury, mutilation or cruelty.
- Hate speech, including content attacking people based on protected characteristics or inciting hatred or genocide.
- Child sexual abuse material (CSAM), child sexual exploitation or any sexual content involving anyone under 18, including synthetic depictions.
- Deepfakes or impersonation of real people: face swaps, nudification, fabricated kissing, hugging or other intimate interactions, or fabricated speech or actions attributed to a real person. These are prohibited even with the person's consent.
- Copyright or trademark infringement, including unauthorized use of protected works, characters, logos or brands.
- Promotion or assistance of terrorism, violent extremism or mass harm, and instructions to make or use dangerous weapons, including weapons of mass destruction.
- Other illegal content, fraud, privacy violations, targeted harassment, instructions to harm people, or threats to create or share prohibited intimate imagery.
- Attempts to bypass content review, including jailbreak prompts, and presenting AI-generated content as authentic footage or human-created work to deceive others.
Use material you have rights to
You must have the necessary rights and permissions for every upload. Photos of identifiable people must be your own likeness or used with their explicit permission; do not upload another person's photo without their consent. Ordinary lawful material depicting people is allowed when it meets these rules, but permission does not exempt any deepfake or impersonation described above. Do not submit confidential information. Model-provider restrictions also apply; generating a video does not guarantee commercial rights.
Automated text checks in the unattended MVP
When the unattended MVP is enabled for a new task, its prompt is sent to Waffo Pancake for an automated text check before submission to the video model provider. Only an allowed text-check result permits generation to start. A check pending confirmation has a fixed time limit and may expire without an allowed result; it is not an operator-approval queue. A blocked, expired, errored or canceled safety check does not start video generation and releases any reserved credits. An inconclusive check is not proof that you violated this policy; try again later if the check could not complete.
Reference images and generated videos, including their audio, are not individually human-reviewed or automatically moderated in this MVP. A passed Waffo Pancake text check does not approve an image or video. Technical delivery verification checks delivery, not content safety, legality, rights or suitability. All the restrictions above still apply to every input and output. You must inspect the result before using or publishing it.
This describes KakaReel's workflow and its limits; it is not a claim that Waffo has approved KakaReel's policy or this change. Model-provider safeguards do not establish that an image or video has been reviewed by KakaReel.
Existing tasks and manual incidents
Existing tasks keep their recorded process. Legacy tasks that require human input or output review still require it, while tasks created before that process retain their earlier delivery rules. We do not label historical tasks as having passed the new text check or as having received a review that never occurred. KakaReel's founder continues to handle legacy reviews, reports and incident investigations.
For paid legacy tasks, credits remain reserved during review. An operator-recorded rejection or expiry of a pre-delivery review releases that reservation; reaching a legacy review deadline alone does not release credits automatically. An unresolved model outcome is a separate issue and may require investigation. Safety checks, reviews and decisions are recorded for accountability.
How we handle violations
- Warning and correction: for a minor or unclear issue, we may explain the concern by email and ask for clarification or a compliant new request. An incident-related hold stays in place while the issue is investigated.
- Hold and investigation: a content, rights or privacy report linked to a task pauses further processing and hides its result in KakaReel while we investigate. A report alone is not proof of a violation.
- Rejection or withheld delivery: when we identify prohibited inputs, we reject them; when we identify prohibited outputs, we withhold them. Routine MVP checks cover prompt text only and may miss violations. Severe issues such as child sexual exploitation, non-consensual intimate imagery, terrorism or credible threats receive priority. We may report suspected unlawful activity through appropriate provider or authority channels when required or appropriate.
- Repeated or serious misuse: the founder can decline further requests after investigating prior cases. This is not an automatic account ban or deletion.
Rejection before delivery releases any credits still reserved for that task, including when the provider has already incurred a cost. That provider cost is KakaReel's responsibility. We do not automatically confiscate unused purchased credits as a penalty. Hiding an already delivered video does not automatically reverse its charge; billing concerns are investigated under our refund and credit rules.
We cannot retract copies already downloaded or guarantee removal from a model provider. We explain the decision and the route to appeal where we can contact the affected user, subject to privacy and legal restrictions. Automatic deletion schedules are not currently implemented; use support for a privacy or deletion request.
Report content or a rights concern
Email support@kakareel.com with “Content report” in the subject, the task ID or affected page if available, a description of the concern, and relevant rights or consent information. You do not need an account or a task ID to report a concern. For child safety, intimate-image abuse, terrorism or a credible threat, start the subject with “Urgent safety report”. Do not attach illegal imagery, passwords, API keys or payment-card details. For immediate danger, contact the appropriate local emergency service; this inbox is not an emergency service.
The founder records the report, assesses its urgency, investigates the available evidence and records the action and reply. We limit access to reporter information to people handling the matter, subject to necessary legal disclosures and our privacy policy. Deliberately false or abusive reports may themselves be investigated; a good-faith report is welcome even if no violation is found.
Response and handling targets
- Urgent safety reports: we aim to provide an initial human response and make an initial hold or withholding decision within 24 hours of receipt. Confirmed severe violations are acted on as soon as they are identified.
- Other content or rights reports: we aim to respond and decide any interim hold within 3 business days, and provide a decision or an investigation update within 7 business days of receipt.
- Appeals: we aim to provide a reviewed decision or a substantive progress update within 7 business days of receipt.
These are operating targets, not guaranteed completion times or a promise of 24/7 staffing. Business days are Monday through Friday, excluding public holidays in China; we use China Standard Time (UTC+8). If we need more time or information, we explain the reason and the next expected update by email. A necessary content hold stays in place while the issue is unresolved. These support targets do not extend a safety check or legacy review deadline. MVP safety-check expiry and legacy operator-recorded expiry follow their respective credit-release rules above.
Appeal a decision
Email support@kakareel.com with “Content appeal” in the subject, the case or task ID if known, the decision you disagree with and any new evidence. Please appeal within 30 days of the notice where possible; this does not limit statutory rights. The founder re-examines the evidence and records a reasoned response. We do not currently have a separate independent appeals reviewer.
Closing a report does not automatically restore a result. An output withheld for an incident or legacy review needs manual review before release. A blocked or expired task needs a new compliant request, with its own quote where applicable and the safety process enabled for that new task; we do not automatically restart generation or charge for it. Credits already released are not charged again merely because an appeal succeeds.