Create broadly.
Within limits we state plainly.

1.Scope

This policy governs everything that moves through Vgent, including anonymous sessions with no account behind them. It sits alongside the Terms of Service and the Privacy Policy. Where a model provider’s own rules are stricter, theirs decide the outcome first.

• Prompts and settings submitted from the landing page composer or the workbench.
• Reference images and extension sources you attach. Uploads are not connected yet; these rules apply from the day they are.
• Videos returned to you, and anything you publish, share, sell or re-edit from them.
• Requests this site makes to its own generation endpoints. The public API and webhooks are not open, so there is no third-party traffic to cover yet.

2.General principles

We do not read prompts before they run and we do not monitor what you make. Three mechanisms carry the weight instead: the model providers' filters at generation time, the hard limits placed on anonymous use, and human review after a report reaches us. Everything below assumes that mix, not a moderation desk watching the queue.

• Generate for purposes that are lawful where you are and where your audience is.
• Do not present a synthetic clip as a recording of something that happened.
• Do not use the service against a real, identifiable person.
• Assume anything you publish outlives the session that made it.

3.What you must not generate

These categories are refused outright. This is not a judgement about taste or genre; it is the line at which a generated clip becomes an instrument of harm. Rewording a prompt does not move that line, and a model accepting the request does not mean the result is permitted here.

• Sexual content involving minors, sexual violence, or any depiction that sexualizes a person who appears to be a minor. Stylized, animated and “aged-up” framings are covered by the same refusal.
• Realistic depictions of a real person being killed, injured or tortured, and anything that instructs, encourages or celebrates suicide, self-harm or disordered eating.
• Terrorist and violent-extremist material: propaganda, recruitment, attack footage, or re-enactments made to glorify an attack or its perpetrators.
• Promotion or instruction for weapons, explosives, controlled substances and other illegal goods, including clips produced to advertise their sale.
• Malware, phishing pages, counterfeit payment or sign-in flows, and video built to run a scam — investment pitches, giveaway bait, impersonated support agents.
• Coordinated harassment, threats, and content that exposes a private individual’s home address, workplace, phone number or other identifying details.
• Deepfakes made to deceive: a real person shown saying or doing something they did not, published without disclosure, or content impersonating an official, an institution or a news outlet.

4.Real likenesses, public figures and synthetic media

A generated face that resembles a real person is that person’s face to everyone watching. Before you publish a clip built on a recognizable likeness you need the right to use it, and your audience needs to know the footage is synthetic.

Public figures are not an exception. Satire and commentary are legitimate uses, but they have to be unmistakable as such, and they must never be framed as a record of a real event, statement or endorsement.

• Obtain documented permission from the person, or from whoever controls their likeness rights, before you publish.
• Label synthetic footage of a real person where you publish it — on the clip itself, not only in a caption that travels separately.
• Never attribute an endorsement, confession, sexual scene or crime to a real person.
• Voice is a likeness. A cloned or imitated voice needs the same permission the face does.

5.Minors

This section has no grey area and no balancing test. We refuse any sexualized depiction of a minor and any content that grooms, endangers or commodifies children, however it is framed — fictional, illustrated, animated, or described as an adult character rendered to look young.

Sessions found generating this material are cut off without notice, and we cooperate with lawful requests from the relevant authorities. The appeals process described below does not apply to this section.

• Do not generate minors in sexual, suggestive or fetish contexts.
• Do not use a real child’s likeness for any purpose without documented consent from their guardian.
• Do not generate violence against children as entertainment.

6.Intellectual property and trademarks

You are responsible for holding the rights to what you generate and publish. A model can produce a character, a logo or a shot that tracks protected work closely, and the fact that a model produced it is not a defence.

The commercial rights to your own outputs are set out in the Terms of Service. Nothing on this page grants you a licence to anyone else’s work.

• Do not reproduce a copyrighted character, film shot or musical work in order to pass it off as your own or as licensed.
• Do not render a brand’s logo, packaging or storefront so that it suggests the brand made, sponsored or approved the clip.
• Reference material has to be yours to use. That condition already holds for the uploads we are still connecting.
• If a rights holder contests a result, we act on the report rather than on who generated it first.

7.Misleading content, news and elections

Generated video is convincing enough that news and civic uses need a higher bar than “technically allowed”. Reconstruction and illustration are fine when they are labelled; fabrication presented as record is not, and the harm does not depend on your intent.

• Do not present generated footage as documentary evidence, breaking news or an eyewitness recording.
• Do not fabricate statements by candidates, officials or election authorities, or footage of voting, fraud or unrest.
• Do not generate false claims about when, where or how to vote, or about a public health or safety emergency.
• When you illustrate a real story with generated footage, disclose it on the clip, not only in a description that can be stripped on re-upload.

8.Model providers apply their own policies

Every generation runs on a third-party model, and each provider enforces its own safety policy at request time. Those policies are stricter than this one in places, looser in others, and they differ from each other. A request this policy permits can still be refused upstream.

When that happens we show you the provider’s refusal instead of inventing a reason of our own: “The request was rejected by the provider’s content filter. Please revise your prompt or references.” We cannot override that decision and we do not appeal it on your behalf.

• The refusal comes back as a content-filter result, not a system failure. Re-submitting the identical prompt will not change it.
• Models draw the line in different places. A prompt one model refuses may run on another, and that is not permission from us.
• Provider policies change without notice to us. Treat this page as our floor, not as a guarantee of what will run.

9.How this is enforced

Enforcement is automatic refusal at generation time plus human review after a report. Because accounts are not open yet, we act on requests and sessions rather than on user accounts, which is a real limit on what enforcement can reach — we would rather say that than imply a system we have not built.

• Provider filters refuse requests as they run. That refusal is immediate and final for the request.
• Anonymous use is capped: a ceiling on the provider cost of any single request, at most 5 generation requests per ten minutes from one network address — visitors behind a shared network draw on the same allowance — and a daily provider budget shared by all anonymous traffic. The 200-credit free allowance quoted elsewhere is a display figure, not a per-visitor balance.
• Reported material is reviewed by a person. Confirmed violations lead to blocked requests, withdrawn access for the session, and removal of anything we host.
• What we retain in order to act on a report, and for how long, is described in the Privacy Policy.

10.Appeals

If you believe a refusal or an enforcement action was wrong, write to us and a person will look at it. Be specific — we cannot review a decision we cannot identify.

A provider’s content-filter refusal is not ours to reverse. What we can do is check whether our own limits were involved, correct our side when they were, and give you a plain answer when they were not. There is no staffed support queue yet, so replies are best-effort rather than guaranteed.

• Include the time of the request, the model you used, and the prompt or settings involved.
• Say what you were making and why you believe it falls inside this policy.
• Expect a direct answer, including when the answer is a refusal we cannot change.
• Appeals do not apply to the minors section. Those decisions are final.

11.Reporting a violation

If you find material generated on Vgent that breaks this policy — above all anything involving a minor, a non-consensual likeness or a credible threat — write to abuse@vgent.com. Include a link or the clip itself, what you believe it violates, and where you encountered it.

If someone is in immediate danger, contact your local emergency services first. We are not an emergency service and cannot act at that speed.

• Reporting a likeness of yourself: tell us how you are identifiable in the clip.
• Reporting copyright: send the work you hold rights to alongside the material you say infringes it.
• Material that appears illegal is referred to the relevant authorities where the law requires it.