AI Content Standards & Safety
Last updated: November 2026
AI-generated video can look very real. These standards set boundaries so the technology is used for creation, not for deceiving, insulting or harming others. They apply together with the Terms of Service.
1. General principles
- Respect real people: do not use anyone else’s image, voice or identity without their permission.
- Be honest with viewers: do not create content that makes viewers believe an event happened when it did not.
- Comply with Vietnamese law, including the rules on advertising, intellectual property and cybersecurity.
2. Absolutely prohibited content
You must not use SONNESTRA to create the following content:
- Violent, terrorist or torture imagery, or content inciting hatred on the basis of ethnicity, religion, gender or sexual orientation.
- Pornographic or obscene content, or any content involving the sexual abuse of children, including simulated imagery.
- Deepfake videos of a person’s identity to scam, defame or extort.
- Misinformation aimed at opposing the State or distorting Vietnamese law.
- Instructions for, or promotion of, self-harm, suicide, illegal gambling, or drug or weapon trafficking.
- Content impersonating government agencies, financial institutions or other brands to deceive viewers.
3. Content requiring caution
The content below is permitted under certain conditions, but you are responsible for meeting those conditions:
| Content type | Conditions to meet |
|---|---|
| Advertising products or services | Stated benefits must be grounded in fact and comply with Vietnamese advertising regulations |
| Medical products, supplements, cosmetics | Do not promise medical cures. Advertise only when the product is licensed and compliant with regulations |
| Finance and investment | Do not guarantee profits. Include clear risk warnings and do not impersonate financial institutions |
| Real people or real events | Do not create the impression that the video is news footage or a genuine statement by that person |
| Educational and historical content | State sources clearly and do not distort verified events |
4. Real people and reference images
When you upload a reference image of a real person, you must have the right to use that person’s image, or have their clear consent for the purpose of creating a video. Reference images may only be used for the job you submitted and must not be shared publicly without the right to do so.
We do not create videos that imitate the voice or face of celebrities or other people to produce statements they never made.
[To be completed by the site owner: the specific policy for images of public figures and celebrities.]
5. Automated moderation
Every prompt and uploaded image passes through a multi-layer AI safety filter before entering the GPU queue. The filter may block by mistake. If your job is blocked, you can rephrase the prompt or use a different image and try again.
Jobs blocked for violations are not refunded. See also the Refund policy.
6. Handling violations
Depending on severity, we apply one of the following measures:
| Level | Example | Measure |
|---|---|---|
| Minor | Ambiguous prompt that may lead to inappropriate content | Block the job and guide the user to edit it |
| Moderate | Repeated minor violations after a warning | Account warning, temporary feature restrictions |
| Severe | Scam deepfakes, or prohibited content listed in section 2 | Temporary suspension or permanent ban of the account; remaining credits are not refunded |
[To be completed by the site owner: warning levels, the length of temporary suspensions, and the procedure for notifying users before a measure is applied. The table above is a proposal that needs to be confirmed.]
In serious violations involving security or causing harm to others, we may retain evidence and cooperate with competent authorities as required by law.
[To be completed by the site owner: the procedure for reporting to cybersecurity authorities when a serious violation is detected.]
7. Reports and appeals
If you see content that violates these rules, created by someone else, or your own content is used without permission, email hotro@sonnestra.vn with the link or job code and a description of the violation. We do not disclose the identity of the reporter to the person reported.
If your job is blocked by mistake, you may submit an appeal with the job code and an explanation. We review appeals manually and reply by email.
[To be completed by the site owner: response time for reports and appeals.]

