Help

Moderation and restrictions

What Problembo moderation checks, what earns points, restrictions and bans, how long they last, and how to write requests so you don't get penalized.

What moderation checks

Before a task starts, Problembo checks the prompt and the uploaded images and videos in image generation, image editing, and video generation. Messages to the Problembot AI agent and comments in the community gallery are checked too. Other tasks are not checked before they start, but a provider's safety refusal in them, such as in music generation, may count as a violation too.

One rule applies to every model, 18+ models included: no sexual content involving minors or people who look underage. Besides that, each model has its own limits on explicit content, and model providers refuse requests under their own safety rules.

Violations count together across all Problembo services: the website, the Telegram bot, the API, MCP, and the Problembot AI agent.

When a refusal counts as a violation

RefusalViolation
Sexual content involving a minorsevere, 5 points
Possibly a minor: the person in the photo looks young, the prompt says "girl", or asks for a younger look1 point
The model's provider refused under its safety rules, and our check finds a violation toousually 4 points; 1 point for some models
A request for help with hacking or illegal acts in a message to the AI agent1 point
Insults, threats, or other harmful text in a comment1 point; the comment is visible only to you
A violation found by a moderator1 or 5 points; for a gross violation, a restriction or ban
The content is too explicit for the chosen modelnot a violation, just a refusal
The provider refused because of a known character, a brand, or another reason that breaks none of our rulesnot a violation
The check is temporarily unavailablenot a violation

The same request repeated within a day is not counted again. Each new wording of a refused request, however, is a new violation.

A refused task does not cost anything: if our moderation refuses it, the task does not start; if the provider refuses it, the reserved money is returned automatically.

What the refusal says

The refusal text tells what happened to your account:

  • how many points the violation added and how many you have out of 10;
  • that the request was not counted again, because it repeats an earlier one;
  • a warning that the next violation like this one will restrict creating content;
  • that creating content is now restricted, and for how long;
  • that the account is banned.

"The safety check is temporarily unavailable" is not a violation: the charge is cancelled, try again in a few minutes.

Points and restrictions

Points count for 24 hours after a violation. When 10 points add up within 24 hours, creating content is restricted, and the count starts again from zero. So a second severe violation within a day already leads to a restriction. The service warns you first: when the next violation like this one would restrict you, the refusal text says so.

Each restriction raises your level by one, and the next restriction lasts longer:

RestrictionDuration
1st30 minutes
2nd3 hours
3rd12 hours
4th3 days
5thpermanent ban

The level drops by one for every 90 days without violations, so the next restriction is short again. A restriction ends on its own when its time is up; support does not lift it early unless the violation was a mistake.

For a gross violation, a moderator may restrict or ban the account at once, without earlier restrictions.

While creating content is restricted

You can't start new tasks on the website, in the Telegram bot, through the API or MCP, message the AI agent, publish works in the community, or leave comments. API requests are refused with the MODERATION_RESTRICTED error and the time the restriction ends.

Signing in, the task history, files, payments, and support stay available, and tasks already started are completed.

Permanent ban

A banned account is disabled: signing in shows that the account is disabled, API tokens and MCP connections stop working. The remaining paid balance is refunded on request to support; the bonus balance is cancelled.

Your moderation status

The Moderation page of your account shows:

  • your points out of 10 and when the nearest ones stop counting;
  • your level, the next restriction, and when the level drops;
  • the restriction in effect and how long it lasts;
  • the scale of restrictions by level;
  • your violations of the last year: date, points, reason, service, the start of the request, the task, and the status: "counts until…", "no longer counts", or "dismissed by a moderator".

How to avoid penalties

  • Make it clear that people are adults. Write "an adult woman", "a 30-year-old man". In explicit requests, avoid words that may mean a child or a teenager: "girl", "schoolgirl", "little", "teen", "loli", school uniform. Don't ask to make a person look younger.
  • Don't upload photos of people who look young for explicit edits, even if they are over 18: moderation judges by how the person looks in the image.
  • Use 18+ models for explicit content. Other models don't accept it, and their providers refuse under their own rules, which may cost points. 18+ models are marked with the 18+ badge in the model list of a service, and the 18+ filter there shows only them. They become available once you enter your birth year in your profile.
  • After a refusal, change the content, not the wording. A rephrased refused request counts as a new violation, so trying different words is riskier than dropping the idea.
  • Stop after a warning. It means the next violation like this one will restrict your account.

If a refusal is a mistake

On the Moderation page, copy the violation ID and contact support. If the mistake is confirmed, the violation is dismissed and stops counting, and a restriction or ban it led to is lifted. A banned account can't open the page: write to support from the account email.

The full rules are in section 4 of the Terms of Use.