Where does the AI generation actually happen?

Inference is performed by ByteDance on Volcengine Ark infrastructure. We send your prompt or image to the Volcengine Ark API and receive the generated clip. We do not run the model ourselves and do not host the model weights.

Are videos I generate watermarked or labeled as AI?

Not at this time. Seedance 2.0 paid output does not embed a visible watermark, and the API does not currently expose C2PA Content Credentials. We are evaluating C2PA adoption as a downstream signing layer. Until that ships, you should label AI-generated content honestly when republishing — the EU AI Act Article 50 imposes this obligation on the deployer.

What happens if I generate something prohibited?

Two layers of enforcement. First, the model-side classifier (operated by ByteDance) rejects many prohibited categories at inference and returns an error rather than the requested clip. Second, our terms prohibit a longer list of operator-side categories — see "Content moderation". Verified violations result in removal of stored output and account suspension or termination.

Can I use videos generated here in commercial work?

Yes, on any paid plan (Basic Pack, monthly, or annual subscription). The grant of commercial license is described in our Terms of Service. Note that commercial license does not equal copyright — copyright eligibility on AI output varies by jurisdiction. For high-stakes commercial uses, consult counsel for your specific jurisdiction.

How do I file a privacy request (DSAR)?

Write to privacy@seedance2-video.com from the email address on the account. State which right you are exercising (access, deletion, portability, etc.). We respond within 30 days. If you are not satisfied with our response, you may complain to your local data protection authority (e.g., ICO in the UK, CPPA in California, or your national DPA in the EU).

How do I report illegal content I see on this service?

Use the channel matching the category — see the "Reporting illegal content" section above. For child safety, write to safety@seedance2-video.com (24-hour acknowledgement, immediate removal for verified reports). For copyright, dmca@. For EU DSA notices, dsa@. For everything else, support@.

Do you transfer data outside the EU / UK?

Yes. Vividra Labs is a US-based controller, and inference is provided by Volcengine Ark which has infrastructure outside the EU. We rely on Standard Contractual Clauses and equivalent transfer mechanisms where applicable. Specifics are in the Privacy Policy.

Is the operator a Delaware shell company?

Vividra Labs LLC is organized in Delaware (a common jurisdiction for US LLCs) with a registered office at 8 The Green, Suite 18, Dover, DE 19901. The operating team works internationally; "Delaware" describes the legal entity's state of formation, not the location of operations. For service of legal process, use the Delaware registered office.

Is Seedance2Video the same as Seedance 2.0 by ByteDance?

No. Seedance 2.0 is the model, built by ByteDance. Seedance2Video is an independent operator (Vividra Labs LLC) that provides a hosted English-language interface to that model via the public Volcengine Ark API.

Are videos from this service AI-generated?

Yes. Every video produced through Seedance2Video is synthetic media generated by the Seedance 2.0 model from your prompt or reference image. This is disclosed throughout the site and is required to be disclosed by the user when republishing under EU AI Act Article 50.

Do you train AI models on my prompts?

No. We do not train, fine-tune, or evaluate any model using your prompts, reference images, or generated videos. We are an operator, not a model developer. The inference itself is provided by Volcengine Ark; their API terms govern that side of the relationship.

Can I use Seedance2Video in the EU?

Yes. We comply with GDPR (data subject rights), the EU AI Act Article 50 (synthetic content transparency), and the EU DSA (notice-and-action mechanism). EU residents have the same rights as US residents, exercised via privacy@seedance2-video.com.

Does this AI transparency page replace the Terms of Service, Privacy Policy, or legal advice?

No. This page summarizes how the service operates for transparency and education. The Terms of Service and Privacy Policy remain the binding legal documents. For specific legal questions, consult qualified counsel in your jurisdiction.

AI transparency disclosure

Our AI transparency commitments

This page is a single-source disclosure of who operates Seedance2Video, what AI model we provide access to, how we handle your data, what is and is not permitted on the service, and how the service maps to the AI transparency obligations of the EU AI Act, GDPR, the EU Digital Services Act, the UK Online Safety Act, US FTC AI guidance, Japan APPI, and Canada PIPEDA. It is written in plain language and does not replace the Terms of Service or Privacy Policy — but it is the first place to look if you want a quick, honest read on how the service runs.

By Jay Yang·14 min read·Last reviewed May 8, 2026·Next review August 8, 2026

Read the disclosure statement Skip to regional compliance

Disclosure statement

In one paragraph

Seedance2Video is operated by Vividra Labs LLC (Dover, Delaware, USA). The video output produced through this service is generated by ByteDance's Seedance 2.0 model, accessed via the Volcengine Ark API. We are an independent third-party operator — not affiliated with, endorsed by, or controlled by ByteDance. Every video generated on this service is synthetic media: it is created by a generative AI system from the user's text or image inputs, and is not a recording of a real event. We do not train, fine-tune, or otherwise modify the underlying model.

All video output on this service is AI-generated synthetic media.
The model (Seedance 2.0) is built by ByteDance; we operate the hosted interface around it.
We pass your prompt or reference image to the Volcengine Ark API and return the result; we do not train on your inputs or outputs.
We comply with applicable laws in the jurisdictions where we operate; see the regional compliance section below for specifics.

Who we are

Operator identity

Seedance2Video is a trade name of Vividra Labs LLC, a limited liability company organized under the laws of Delaware, USA, with a registered office at 8 The Green, Suite 18, Dover, DE 19901. The website seedance2-video.com and the related localized pages under /zh, /es, /ko and other locales are the only domains operated by Vividra Labs LLC for this service.

We are not ByteDance, not Volcengine, not Doubao, and not Jimeng. Those are separately operated services and brands of ByteDance Ltd. We provide a hosted English-language interface that calls the public Seedance 2.0 API on behalf of registered users. The model output you see on this site is identical to what the same prompt would return through Volcengine Ark or any other compliant integrator; the value we add is the interface, account system, credit-based billing, and global payment support.

The model

Seedance 2.0 — model provenance

Every video produced on this service is generated by Seedance 2.0. The fields below describe the model and the API endpoint we call. They are sourced from ByteDance Seed and Volcengine Ark public documentation.

Model name: Seedance 2.0
Vendor: ByteDance Ltd.
Brand family: Doubao (豆包)
Modalities: Text-to-video and image-to-video
Released: February 12, 2026
API public beta: Volcengine Ark — April 2026
Max output: Up to 12 seconds at 1080p with native audio
API endpoint we call: Volcengine Ark Seedance 2.0 inference API
Modifications by us: None. We do not fine-tune, distill, or otherwise modify the model.

Our role

What Seedance2Video does and does not do

We do four things: (1) provide a web interface where you write a prompt or upload a reference image, (2) authenticate your account and meter credit usage, (3) call the Seedance 2.0 API on Volcengine Ark with your input, and (4) return the model output and store it in your account library for thirty days so you can download it.

We do not train, fine-tune, distill, quantize, or otherwise alter the underlying model. We do not use your prompts or generated videos to train any model — ours or third-party. We do not sell your inputs or outputs to data brokers. We do not share your account information with advertisers. The model and the data flow are described under "Data handling" below.

Data

How your data flows through this service

When you generate a video, your input passes through three systems: our application (Vividra Labs), our subprocessors (payment, hosting, analytics), and ByteDance / Volcengine Ark for inference. Each row below states what data each system handles and how long it is retained.

Account data: Email, hashed password (or OAuth token), and credit balance are stored on our infrastructure. Retained for the life of the account; deleted within 30 days of account closure or DSAR-driven request.
Prompt text and reference images: Sent in real time to Volcengine Ark for inference. Stored in your account history on our side so you can re-render or review. We do not use prompts or images to train any model. Retention: 30 days for output media, indefinite for the prompt-and-seed record (so you can re-run a winning prompt).
Generated videos: Stored in your My Videos library on our cloud storage for 30 days from generation. After 30 days the file is deleted; the prompt-and-seed record remains so you can regenerate.
Payment data: Processed by Stripe (and PayPal where elected). We do not store full card numbers. PCI scope is delegated to the processor.
Analytics: Aggregate, privacy-respecting analytics for traffic and conversion. No fingerprinting; no advertising trackers on this site at the time of writing.
Server logs: Standard request logs (IP, user agent, timestamp) retained for security and abuse-investigation purposes for 90 days.

Training data

We do not train on your content

Vividra Labs does not operate a model and does not have a training pipeline. We will not use your prompts, your reference images, or the videos generated for you to train, fine-tune, or evaluate any AI system. This commitment is binding regardless of plan tier.

Inference is provided by ByteDance via the Volcengine Ark API. Their API terms govern what they may do with API requests; consult Volcengine Ark documentation directly for their position. We are not a party to that side of the relationship and cannot speak for it. If you have concerns about the model-side handling of your content, you should evaluate Volcengine Ark's terms in addition to ours before generating sensitive material.

Moderation

Prohibited content and enforcement

The Seedance 2.0 API enforces ByteDance's safety classifiers at the inference layer — generations that match flagged categories are rejected at the model side and never reach your screen. We add a layer of policy on the operator side covering use-cases where the model output is technically generable but the use is illegal or harmful. The categories below are prohibited on Seedance2Video.

Sexual content involving minors: Absolute prohibition. Reports to NCMEC where applicable. Account terminated, no refund.
Non-consensual intimate imagery: Including deepfakes of real people. Removed and account suspended on first verified report.
Identifiable persons without consent: Using the likeness of a real, identifiable private person without their consent — including political deepfakes and impersonation — is prohibited.
Violent extremist content: Content promoting terrorism or violent extremist organizations (per UN designations). Removed and reported to authorities where required.
Disinformation in regulated domains: Synthetic content presenting as authentic news, election material, or medical/financial advice. Prohibited under our terms; may also be illegal under local law.
Child-endangering content: Content depicting minors in sexual, dangerous, or violent contexts. Absolute prohibition with mandatory reporting.
Targeted harassment: Content created to harass, threaten, or doxx an identifiable person. Removed; repeat offenders banned.

Reports of prohibited content can be sent to support@seedance2-video.com. We acknowledge within 48 hours and act within 7 days for verified reports — faster for child-safety and immediate-harm categories. Enforcement actions include removal of stored generations, account suspension or termination, and where applicable referral to law enforcement.

Provenance

Watermarking and content credentials

As of May 2026, Seedance 2.0 does not embed visible watermarks on output for paid plans, and the Volcengine Ark API does not currently expose C2PA Content Credentials in the returned media. This means downstream tools cannot automatically verify that a clip from this service is AI-generated unless you explicitly label it as such.

We are evaluating C2PA Content Credentials adoption (the open standard from the Coalition for Content Provenance and Authenticity) as a downstream layer we can apply at file delivery. Until that is shipped, we ask all users to label AI-generated content honestly when publishing — both because it is the right thing to do and because the EU AI Act Article 50 imposes legal disclosure obligations on the deployer of generative AI systems for content that could be mistaken for authentic.

When C2PA support ships on this service, this section will be updated and the dateModified field will reflect the change.

Limitations

What Seedance 2.0 is and is not good at

Generative video models — including Seedance 2.0 — have systematic weaknesses that users should understand before relying on the output for high-stakes purposes.

Bias: Like all large generative models, Seedance 2.0 reflects biases present in its training corpus. Generic prompts ("a doctor", "a CEO") often default to stereotyped depictions; we recommend explicit subject descriptions to override defaults.
Long coherence: Identity and style drift across clips longer than about 8 seconds, and across separate generations. Multi-shot sequences are best assembled in a video editor from individual clips with locked seeds.
On-screen text: In-frame typography is unreliable. For accurate text in videos, generate silent footage and overlay text in post-production.
Real-world physics: Physics violations (objects passing through each other, water behaving wrongly) appear in a meaningful percentage of generations. Always review output before publishing.
Identifiable person likeness: The model can generate realistic human faces. Generating the likeness of a specific real, identifiable person is prohibited by our policy and by the model-side classifier — but accidental resemblance to a real person is possible. Review your output and re-prompt if needed.

Your rights

What you can ask us to do

Different jurisdictions grant different rights over personal data. We honor the strongest applicable rights for every user regardless of residency, with the regional reference linked below for users who want the legal text. To exercise any right listed here, write to privacy@seedance2-video.com from the email address on the account.

Access: You can request a copy of all personal data we hold about you, including account record, generation history, and stored videos. We respond within 30 days (GDPR Art. 15 / CCPA §1798.110).
Deletion / erasure: You can request deletion of your account and associated data. We delete within 30 days, except where retention is required by law or for fraud prevention (GDPR Art. 17 / CCPA §1798.105).
Portability: You can request a machine-readable export of your prompts and generated videos. We provide this within 30 days as a downloadable archive (GDPR Art. 20).
Correction / rectification: You can correct inaccurate personal data on your account at any time from settings, or by writing to privacy@ for fields you cannot edit yourself (GDPR Art. 16).
Objection: You can object to processing for any specific purpose. Where processing is required to deliver the service, we will explain the requirement; for optional processing we will stop (GDPR Art. 21).
Opt-out of sale / sharing: We do not sell or share personal data within the meaning of CCPA / CPRA. There is no opt-out form needed because there is nothing to opt out of.
Complaint: If we do not address a request to your satisfaction, you may complain to your local data protection authority — for EU residents, your national DPA; for UK residents, the ICO; for California residents, the CPPA.

Reporting

Notice & action — how to report illegal or harmful content

We provide a notice-and-action mechanism aligned with EU DSA Articles 16–17 and the UK Online Safety Act. The channels below are open to any person, regardless of whether they are a Seedance2Video user. We acknowledge every report and notify the reporter of the outcome within the SLA listed.

Category	Email	SLA
General prohibited content	support@seedance2-video.com	Acknowledgement within 48 hours; action within 7 days.
Child safety / CSAM	safety@seedance2-video.com	Acknowledgement within 24 hours; immediate removal and mandatory reporting (NCMEC where applicable).
Copyright (DMCA — US)	dmca@seedance2-video.com	Acknowledgement within 48 hours; action within 10 business days for compliant notices.
EU DSA notices	dsa@seedance2-video.com	Acknowledgement within 48 hours; statement of reasons issued for any action.
Privacy / data subject requests	privacy@seedance2-video.com	Response within 30 days (GDPR / CCPA standard).

Regional compliance

Compliance position by jurisdiction

This is a non-exhaustive map of the principal AI / privacy / online-safety frameworks in the markets where we have meaningful traffic, and where Seedance2Video stands relative to each. Where a framework applies, we comply with it; where compliance is in progress, we say so.

Region	Framework	Applicability	Our position
🇪🇺 EU	EU AI Act (Reg. 2024/1689)	Article 50 transparency obligations apply to deployers of generative AI: synthetic content disclosure required.	Disclosed throughout this site that all output is AI-generated. C2PA Content Credentials evaluation in progress.
🇪🇺 EU	GDPR (Reg. 2016/679)	Personal data of EU residents — full data-subject rights under Articles 15–22.	Honored. Single contact at privacy@. Data flows documented above.
🇪🇺 EU	EU Digital Services Act (Reg. 2022/2065)	Notice-and-action mechanism, statement of reasons, complaint handling.	Implemented. Dedicated dsa@ channel; statements of reasons issued for moderation actions.
🇬🇧 UK	Online Safety Act 2023	Illegal-content duty for user-to-user services. Children's safety duties where the service is likely to be accessed by children.	Illegal-content reporting channel live; minimum age 13 enforced at signup. Children's risk assessment in progress.
🇺🇸 US	FTC AI guidance (Section 5 FTC Act)	Prohibition on unfair or deceptive AI practices; honest material disclosures.	Honest disclosures throughout: operator identity, model provenance, no free-trial misdirection, plain-language policies.
🇺🇸 California	CCPA / CPRA (Cal. Civ. Code §1798.100 et seq.)	Notice at collection, consumer rights including opt-out of sale/sharing.	No sale or sharing of personal data in CCPA scope. Rights honored via privacy@.
🇨🇦 Canada	PIPEDA	Consent-based collection, use, and disclosure of personal information by commercial organizations.	Consent obtained at signup. Rights honored via privacy@.
🇯🇵 Japan	APPI (Act on the Protection of Personal Information)	Cross-border transfers and data subject rights for personal information of Japanese residents.	Honored. Cross-border transfer disclosed (US-based Vividra Labs as controller; Volcengine for inference).
🇦🇺 Australia	Privacy Act 1988 (APPs)	Australian Privacy Principles for handling personal information.	Honored. Data flows align with APP 1, 6, 8. Cross-border disclosure documented.

Glossary

Compliance terms in plain language

These are the abbreviations and technical terms used on this page. Each definition is a one-paragraph plain-language summary, not a substitute for the legal text.

EU AI Act ↗: Regulation (EU) 2024/1689 governing artificial intelligence in the European Union. Article 50 imposes transparency obligations on deployers of generative AI: users must be informed they are interacting with AI, and synthetic content that could be mistaken for authentic must be marked as artificially generated.
GDPR ↗: General Data Protection Regulation — Regulation (EU) 2016/679. Governs the processing of personal data of individuals in the EU. Confers rights of access, erasure, portability, and objection on data subjects.
EU DSA ↗: Digital Services Act — Regulation (EU) 2022/2065. Imposes content-moderation, transparency, and notice-and-action duties on hosting and platform services that operate in the EU.
UK OSA ↗: United Kingdom Online Safety Act 2023. Imposes illegal-content and children's safety duties on user-to-user and search services with UK users.
FTC ↗: United States Federal Trade Commission. Enforces Section 5 of the FTC Act against unfair or deceptive business practices, including in AI products.
CCPA / CPRA ↗: California Consumer Privacy Act, as amended by the California Privacy Rights Act. Gives California residents rights of access, deletion, correction, and opt-out of sale/sharing of personal information.
APPI ↗: Japan's Act on the Protection of Personal Information. The principal Japanese privacy framework, with specific rules for cross-border transfer of personal data of Japanese residents.
PIPEDA ↗: Canada's Personal Information Protection and Electronic Documents Act. Governs how private-sector organizations collect, use, and disclose personal information in the course of commercial activity.
C2PA Content Credentials ↗: An open standard from the Coalition for Content Provenance and Authenticity for cryptographically signing media with provenance metadata, including whether the asset was AI-generated. We are evaluating adoption.
NCMEC ↗: National Center for Missing & Exploited Children — the US-based clearinghouse for child sexual abuse material reports. Mandatory reporting destination for US-based providers under 18 U.S.C. §2258A.

Verification

How we keep this page honest

This page is reviewed quarterly by our editorial team and updated on the dateModified field whenever a control changes — for example, if a new subprocessor is added, if C2PA Content Credentials are implemented, if a new jurisdiction enters scope, or if a regulatory framework changes the obligations described here.

When you read a control statement on this page (e.g. "we do not train on your inputs", "C2PA Content Credentials are not currently embedded"), you can take it as a description of the operating reality on the dateModified date. If you observe something inconsistent with the statement, please write to support@seedance2-video.com — we treat documentation drift as a high-severity issue.

Document version: 2026-05-08·Next scheduled review: 2026-08-08

Quick answers

Is Seedance2Video the same as Seedance 2.0 by ByteDance?: No. Seedance 2.0 is the model, built by ByteDance. Seedance2Video is an independent operator (Vividra Labs LLC) that provides a hosted English-language interface to that model via the public Volcengine Ark API.
Are videos from this service AI-generated?: Yes. Every video produced through Seedance2Video is synthetic media generated by the Seedance 2.0 model from your prompt or reference image. This is disclosed throughout the site and is required to be disclosed by the user when republishing under EU AI Act Article 50.
Do you train AI models on my prompts?: No. We do not train, fine-tune, or evaluate any model using your prompts, reference images, or generated videos. We are an operator, not a model developer. The inference itself is provided by Volcengine Ark; their API terms govern that side of the relationship.
Can I use Seedance2Video in the EU?: Yes. We comply with GDPR (data subject rights), the EU AI Act Article 50 (synthetic content transparency), and the EU DSA (notice-and-action mechanism). EU residents have the same rights as US residents, exercised via privacy@seedance2-video.com.

FAQ

Frequently asked questions

Where does the AI generation actually happen?: Inference is performed by ByteDance on Volcengine Ark infrastructure. We send your prompt or image to the Volcengine Ark API and receive the generated clip. We do not run the model ourselves and do not host the model weights.
Are videos I generate watermarked or labeled as AI?: Not at this time. Seedance 2.0 paid output does not embed a visible watermark, and the API does not currently expose C2PA Content Credentials. We are evaluating C2PA adoption as a downstream signing layer. Until that ships, you should label AI-generated content honestly when republishing — the EU AI Act Article 50 imposes this obligation on the deployer.
What happens if I generate something prohibited?: Two layers of enforcement. First, the model-side classifier (operated by ByteDance) rejects many prohibited categories at inference and returns an error rather than the requested clip. Second, our terms prohibit a longer list of operator-side categories — see "Content moderation". Verified violations result in removal of stored output and account suspension or termination.
Can I use videos generated here in commercial work?: Yes, on any paid plan (Basic Pack, monthly, or annual subscription). The grant of commercial license is described in our Terms of Service. Note that commercial license does not equal copyright — copyright eligibility on AI output varies by jurisdiction. For high-stakes commercial uses, consult counsel for your specific jurisdiction.
How do I file a privacy request (DSAR)?: Write to privacy@seedance2-video.com from the email address on the account. State which right you are exercising (access, deletion, portability, etc.). We respond within 30 days. If you are not satisfied with our response, you may complain to your local data protection authority (e.g., ICO in the UK, CPPA in California, or your national DPA in the EU).
How do I report illegal content I see on this service?: Use the channel matching the category — see the "Reporting illegal content" section above. For child safety, write to safety@seedance2-video.com (24-hour acknowledgement, immediate removal for verified reports). For copyright, dmca@. For EU DSA notices, dsa@. For everything else, support@.
Do you transfer data outside the EU / UK?: Yes. Vividra Labs is a US-based controller, and inference is provided by Volcengine Ark which has infrastructure outside the EU. We rely on Standard Contractual Clauses and equivalent transfer mechanisms where applicable. Specifics are in the Privacy Policy.
Is the operator a Delaware shell company?: Vividra Labs LLC is organized in Delaware (a common jurisdiction for US LLCs) with a registered office at 8 The Green, Suite 18, Dover, DE 19901. The operating team works internationally; "Delaware" describes the legal entity's state of formation, not the location of operations. For service of legal process, use the Delaware registered office.