changelog-graphic-v7
You can now report a message or a person and block someone in two taps: the talk ends for both of you without saying who ended it, and you will not be matched again. If a message points to a crisis or to violence, the person concerned privately receives help numbers and the AI switches to a calm, protective mode. Under Sources, Wikipedia now shows up to five matching articles so you can pick the right one, and every mediator message carries an AI badge.
Safety
  • Report.
    On a message from the other person or from the AI, tap ⋯ and choose Report message; on a profile, Report person. Pick one of seven reasons and add a note if you like. A message you reported disappears for you right away, and we look at every report within 24 hours.
  • Block.
    On a profile: Block person. The talk ends for both of you without a hint who ended it, open invitations between you are cancelled, and you will not be matched or invited again. Unblock any time under Settings → Account → Blocked people. Unblocking does not reopen ended talks.
  • Screening in talks with strangers.
    In talks that started from the topic cards, the AI checks each new message for serious violations only: credible threats, hate against groups, sexual content, anything involving minors, sharing private data. Such a message is not shown to the other person, the writer gets a private note, and our moderation is informed. Harsh words in an argument are not filtered, and talks by invitation are not screened this way.
  • Support when it matters.
    If a message points to a crisis, or to violence or coercion, the person concerned receives help numbers in a private system message, once per talk, in the app language. The AI then stops working toward an agreement: it stays calm and neutral, offers a pause and never names the trigger.
  • Rules and 18+.
    When you create an account you now accept the rules (no harassment, threats, hate or sexual content; we may hide content, end talks and suspend accounts) and confirm that you are at least 18. WorldViews is rated 18+ in the App Store and on Google Play.
  • Before you share contact details.
    While you type a phone number, an e-mail address or a street address, a quiet hint reminds you to share it only with someone you trust. WorldViews never asks for it. Sending images in a talk is off for now.
  • Moderation.
    Reports reach us as a push notification and are handled from a moderation screen inside the app, also while travelling. Accounts that break the rules can be suspended; a suspended account can read its talks but not start new ones or send messages.
AI
  • AI badge.
    Every message from the mediator shows a small AI badge next to its name.
  • Named provider.
    The consent text at sign-up now says that talk content is processed by Google Gemini for the AI mediation.
Sources
  • Pick the right Wikipedia article.
    Sources → Wikipedia shows up to five matching articles with title and first lines. Tap one to see the preview; Open original opens exactly that article, Insert quote quotes it, Back to results returns to the list. A new search resets the choice. If a topic word finds little in your language's Wikipedia, use the search line.
Prepare and Cool down
  • The preparation wizard shows your topic's tag word above Step 1 of 3.
  • Breathe for a moment has proper spacing inside its card.
  • The Waiting for a partner label on Connect now hugs its text.
Fixes
  • Icons that showed as boxes in release builds (navigation bar, Wikipedia, quote, Apple sign-in and others) render again.
  • The "no article found" hint under Sources appears when a search has no result.
  • The talk list preview never shows a hidden message.