Local readiness inspection
The browser checks selected file size, dimensions, video duration, and batch totals before upload. This readiness check does not upload the selected media.
Privacy and consent guide
Face swap safety depends on both the service and the intended use. Build a local pre-publication record, confirm who gave permission, and inspect what the provider retains and the finished file reveals before you generate or share it.
Before upload
| Question | Proceed when | Stop or clarify when |
|---|---|---|
| Do you have permission? | Every real person understands the edit and intended audience | A public photo is your only basis for using the likeness |
| Could the result mislead someone? | The creative context is clear or the edit will be disclosed | It could be mistaken for evidence, endorsement, or a real event |
| Is the context sensitive? | The use is lawful, permitted, and does not exploit or harm the subject | It involves minors, sexual content, fraud, harassment, or deceptive impersonation |
| What happens to uploads? | Retention, deletion, training use, and providers are explained | The policy is vague or contradicts the product interface |
| What will the exported file reveal? | You have inspected watermarks and visible account information | You would not want that information shown to the intended audience |
Five-step review
Current DeepSwapAI controls
The browser checks selected file size, dimensions, video duration, and batch totals before upload. This readiness check does not upload the selected media.
When a task is submitted, the supplied media is sent to an AI inference provider only to complete that requested generation.
DeepSwapAI does not use uploaded content to train AI models.
Uploads and generated media are automatically removed from DeepSwapAI servers within 24 hours. Account, order, credit, security, and support records can be retained where needed for service operation, fraud prevention, accounting, disputes, or legal compliance.
A signed-in user can permanently delete the account from account settings. Privacy requests can also be submitted through the official support channels.
Read the current Privacy Policy, Terms of Service, and Trust Center before relying on these controls.
Disclosure decision
| Destination | Minimum documented decision | Escalate when |
|---|---|---|
| Private draft | Permission and asset rights still apply; record who can access the file | The draft could leave the intended group or contains sensitive material |
| Internal or client review | Name the approval owner, permitted reviewers, and deletion path | The client cannot confirm likeness or source-media rights |
| Organic public post | Review whether a viewer could mistake the edit for a real appearance, event, statement, or endorsement | The destination rule, permission scope, or disclosure placement is unclear |
| Paid ad or endorsement | Confirm the approved claim, channel, territory, duration, paid placement, and final-result approval | Any likeness, advertising, contractual, or platform requirement is unresolved |
| Sensitive or high-impact context | Pause and obtain qualified review before generation or publication | The edit touches minors, intimacy, fraud, politics, health, finance, crime, harassment, or identity verification |
Disclosure is not a substitute for permission. A visible label can help a viewer understand that media was altered, but it does not authorize the likeness, source asset, message, or distribution.
Requests and reports
Use account settings or the privacy request channel listed on the official contact page.
Report the exact URL, describe why the subject did not consent or how the content violates the Terms, and avoid sending private source media unless support requests it.
Identify the protected work, the disputed material, and the exact location where it appears.
Provide reproducible steps and a safe proof of concept without accessing, changing, or retaining another user's data.
Official sources and evidence boundary
Consent and disclosure questions
No. Public visibility does not by itself establish permission to alter a likeness, change its context, or distribute the edited result.
No. The checklist is a planning and review record, not a release, signature, identity verification, legal opinion, or proof that permission was valid.
Document a disclosure decision whenever viewers could reasonably mistake the result for a real appearance, statement, event, or endorsement, and check the current rules for the destination and jurisdiction.
No. C2PA Content Credentials can provide tamper-evident provenance assertions, but the C2PA specification says provenance alone does not establish whether content is true, accurate, factual, or properly consented.
No. It has no media input, account request, network request, analytics event, or server save. Entries remain in the current browser tab unless you download a local JSON or CSV record.
How this guide was made
The DeepSwapAI Product Team reviewed the current media-readiness client, generation flow, 24-hour cleanup job, account-deletion endpoint, complete-email watermark entitlement behavior, Privacy Policy, Terms, Trust Center, support routes, local planner code, and the official sources listed above. The planner was designed as a local decision record with explicit evidence limits, not a release or automated legal decision. Product controls and public guidance were reviewed on July 22, 2026; see the verification methodology.
Use one permitted photo for the smallest representative test only after the permission, source-media, sensitive-use, export, and disclosure decisions are complete.