Every genvi generation is now checked before you pay

From today, genvi checks every generation before you pay for it. If a request breaks our rules, it stops before we take any credits, and we tell you which rule in plain words. It costs nothing.

The same rule applies on the web, the phone apps, chat, the agent and the developer tools. We refuse:

Battle scenes and horror still pass. We block graphic gore and threats, not stories with conflict in them.

You get a short message that names the reason. For a prompt, it reads like this: "This prompt can't be generated: it asks for sexual content, which genvi doesn't make." If the problem is a picture you added, the message says that a picture you gave shows that kind of content. We check up to eight pictures you add, and pictures you add with an @mention (typing @ and a name) are included. In chat we check your own words and the plan the assistant writes from them, so a refused message leaves nothing behind.

Because the check happens before we take credits, there is nothing to refund. If you want the engineering story, read moderation before the credit reservation.

We also turned on the safety filters that the companies behind each model offer, where they have one. Sharing to Explore is checked the same way, including your words, the picture and any reference pictures you share. If the check is not available when you share, sharing waits a moment instead of letting a post through.

You can report anything genvi makes, right where it is shown. In the phone apps, Report is in the menu of a chat result, a file in the Library, each agent output, the Image Composer result and a video clip. Pick a reason, such as Violence or gore or Shows a real person without consent, and add a note if you like. The app replies "Thanks. We review reports within 24 hours." Every report also emails our support team.

Nothing. The check is part of every generation and is never charged.