Seedance Output Video Sensitive Content: Failed Tasks, Retries and Refunds
Resolve OutputVideoSensitiveContentDetected: why a Seedance task fails after submission, where to check the error and refund, and when a new generation makes sense.
Published
A submitted Seedance task can later fail output review, even when the input was accepted. Read the task's final error, then its settlement receipt. Do not repeatedly request a download from a failed task or treat HTTP 200 as proof that a usable video exists. A new generation is a separate attempt.
Using Seedance through an API? Compare prices before your next generation. APIMaster offers the four models below, with task-status and billing-receipt endpoints. A cheaper route does not remove content review.
Checked 2026-11-11. USD per generated second at 720p, with generated audio and no reference-video input. Only enabled public routes are counted. The live model card takes precedence.
| Model ID | Reference price | APIMaster current price | Active routes | Discount | Model card |
|---|---|---|---|---|---|
seedance-2.5 |
$0.23112/s | From $0.218917/s | 6 | Up to ~5.3% off reference | View price |
seedance-2.0 |
$0.1512/s | From $0.143217/s | 6 | Up to ~5.3% off reference | View price |
seedance-2.0-fast |
$0.12096/s | From $0.072503/s | 4 | Up to ~40.1% off reference | View price |
seedance-2.0-mini |
$0.0756/s | From $0.037649/s | 4 | Up to ~50.2% off reference | View price |
Reference prices are marketplace configuration references, not independently verified manufacturer checkout prices. Route prices vary; Fast and Mini are different models, not discounted full 2.5. For same-model provider comparisons, see Seedance Best Price.
GPT Trial(GPT-6 Astra, GPT-6 Sol, GPT-5.6, GPT-5.5, gpt-5.6-luna, gpt-5.6-sol, gpt-5.6-terra, gpt-6.1-sol, gpt-6-astra, gpt-6-sol, and gpt-6-luna available)
Sign up and claim a $20 GPT trial
Recognize an output-video rejection
OutputVideoSensitiveContentDetected
The request failed because the output video may contain sensitive information.
This points to review of the generated video, rather than just the text or uploaded image. A clean-looking input does not guarantee an accepted output. The response does not reveal the exact flagged frame, prove that a complete deliverable is stored, or establish a specific false-positive rate.
If the suffix is .PolicyViolation and the message mentions copyright restrictions, use the copyright guide. If it names output audio, inspect the audio branch instead of calling it a video-image problem.
Why submission succeeds but video generation fails
An asynchronous creation response means a task was submitted, not that all generation and review stages passed. Your app should show the final task state rather than leaving a submitted job labeled "success."
For an APIMaster task, retrieve its status using the original public task ID:
curl 'https://apimaster.ai/v1/videos/task_example' \
-H 'Authorization: Bearer YOUR_API_KEY'
Read top-level status and error; the compatibility endpoint can use data.status and data.error. HTTP 200 describes a successful query, not necessarily a successful task. A schematic failed response looks like this:
{
"id": "task_example",
"status": "failed",
"error": {
"code": "OutputVideoSensitiveContentDetected",
"message": "The request failed because the output video may contain sensitive information."
}
}
Actual response fields can vary. Preserve the returned error rather than requiring this exact envelope. Download only from a confirmed completed task with its returned result location; there is no promised public URL for a rejected output.
What should I change before another generation?
Review the entire creative brief and reference media together. Ask whether the requested motion or scene produces content outside the provider's rules, or whether an application added unintended instructions. Make the intended legitimate action explicit and remove unsupported requests. When the creative goal is valid but the reason is unclear, ask for review with the original task ID.
Do not invent a particular flagged frame when you cannot view the rejected output. Avoid treating a broad moderation message as proof that one specific word caused it. Changing a seed, removing names or buying another route does not establish that the content issue has been resolved.
Repeatedly polling the failed task will not generate a different result. Repeated new submissions can also waste time and create more billing records without improving the diagnosis. Keep a record of the change you are testing and its actual outcome.
Does APIMaster refund a failed Seedance video?
Current Seedance billing documentation says confirmed provider-task failures trigger a full generation refund. Verify the receipt rather than assuming the first debit row is the final charge:
curl 'https://apimaster.ai/v1/videos/task_example/billing' \
-H 'Authorization: Bearer YOUR_API_KEY'
| Receipt state | What it tells you | What to do |
|---|---|---|
pending |
Financial processing is not final | Retain the task and check later; escalate an unresolved receipt |
refunded |
Refund has finalized | Confirm refunded_amount and zero net_amount for a full refund |
settled |
Charge has finalized | Compare the final amount with task outcome; ask support about any discrepancy |
| HTTP 404 | No accessible receipt was returned | Check account/task identity and whether it is a historical task; do not infer zero cost |
Task completion and settlement are separate. The receipt endpoint is read-only: querying it does not launch generation or issue another refund. These APIMaster generation rules do not establish another platform's billing policy or cover every independent asset-service fee.
Timeout or missing task: a different recovery problem
A confirmed output failure is different from losing the creation response. If a POST times out, the provider may still have accepted it. For a supported originally keyed submission, recover its saved receipt; do not submit a fresh job just because your browser stopped waiting.
For a confirmed failed task, any corrected replacement is a new generation with a new identity. For uncertain acceptance, reconcile the original attempt first. Keep the original account, API base, payload and key for recovery. A refund or missing task alone does not prove that an uncertain upstream attempt cannot still run.
Use Seedance with visible pricing and task receipts
Check Seedance 2.5 pricing, or compare Fast and Mini for a different budget. Then register, create an API key, and use https://apimaster.ai/v1 with the exact model ID, such as seedance-2.5. Read the generation API guide before submitting a paid request.
For a supported request, obtain a non-binding quote before generation and keep the resulting task ID for status and settlement checks. Switching platforms or paying less does not guarantee moderation approval.
Sources and scope
Error meanings were checked against the official ModelArk error catalog on 2026-11-11 and redacted operational error messages. Positions are normalized to content[n]; request IDs and private media are omitted. Examples illustrate response handling, not a new paid generation test. Provider wording, supported assets and billing rules can differ. APIMaster procedures follow its linked current documentation.
Different error? Open the Seedance error index.
FAQ
Can I download the video that moderation rejected?
There is no guaranteed downloadable result for a failed reviewed task. Use the result location returned for a confirmed completed task, rather than constructing a hidden output URL.
Does turning audio off fix an output-video error?
It is not a diagnosis of the video error. If the actual code names output audio and the project does not require generated sound, a supported audio-off option changes that workflow. Video review still applies.
What should a support ticket include?
The original task/submission ID, model, UTC time, full redacted output error, intended legitimate scene and billing-receipt state. Never include the API key or private access links in a public report.
