AI and your game data

Does Boardssey share your games with generative AI? What we never do with your designs, which features send text to an AI model when you press their button, and how to keep a design out of them.

Last reviewed November 16, 2026

Your unpublished game is your work, and it stays yours. Boardssey does not train AI models on your games, does not sell or hand your data to AI companies, and does not send anything to an AI service in the background. This page is for anyone who wants to know exactly where their game goes before they use a feature.

What this page helps you do

  • Get the short answer to "does Boardssey share my game with AI?"
  • See which features use an AI model, what each one sends, and when.
  • Understand what the AI provider may keep, and for how long.
  • Keep a design completely away from AI, if you prefer.

The short answer

  • No training. Boardssey never uses your games, rules, artwork, playtests or any other team data to train an AI model, and never gives or sells it to anyone who would.
  • Nothing in the background. No AI reads your workspace on its own. A handful of features call an AI model, and each one does it only when you press its button, sending only what that feature needs for that one answer.
  • The providers do not train on it either. Those features use the business APIs of two AI companies: Anthropic (Claude) for most of them, OpenAI for the help bot and meeting dictation. Under both companies' API terms, data sent through the API is not used to train their models.
  • Your artwork stays out. No feature sends your artwork, card images or 3D files to an AI model. The only picture that ever goes is a photo of a business card, when you scan one.

Helpful details

  • Key areas: the features are listed in the next section. If you never press their buttons, nothing from your team reaches an AI model.
  • What you can do: use Boardssey with none of them. Every feature on this page is optional, and the rest of the app works without it.
  • Good to know: Boardssey keeps a count of these requests per team (to apply the daily limits), not what was in them.

Features that use an AI model

Each feature below sends its request to the provider named in the table, and only when you use it.

FeatureWhat is sentWhenProvider
AI commands and the next-words suggestion in the text editor (rules, notes, descriptions, task descriptions)The paragraph or the text you selected, and your instructionWhen you pick an AI command, or press Ctrl+Shift+Space / Cmd+Shift+Space (or Ctrl+Space) for a suggestionAnthropic
Scan Business Card (Contacts)The photo of the cardWhen you scan a cardAnthropic
Dictate in meeting notesThe audio you record, plus the meeting's title and company to help spell namesWhen you click Dictate and stop recordingOpenAI
Auto Translate (catalog translations)The game's title, short description, description, three hooks and component namesWhen you click Auto TranslateAnthropic
IDML TranslatorThe text of the rulebook's paragraphs (and, for the glossary, the document's text), not its images or layoutWhen you translate a fileAnthropic
Quick Setup (Virtual Table)Your description, the names and kinds of the table's components and buttons, and the game's rules only if you tick Use the rules of {game}When you press Quick SetupAnthropic
Quick Scene (3D Mockups)Your description and the scene's layout: kinds of pieces, sizes, positions, camera and lighting. Not your artworkWhen you pick Quick Scene in the Scenes menuAnthropic
Quick Insert (Insert Designer)Your description, the box size and the materialWhen you press Quick InsertAnthropic
Quick Punchboard (Dieline Generator)Your description and the sheet sizeWhen you press Quick PunchboardAnthropic
Help bot (the chat bubble in the bottom right)Your question, the earlier messages of that conversation, and your team's plan, subscription status, your role and which plan features you haveWhen you send a questionOpenAI

Helpful details

  • Key areas: the help bot does not read your games, playtests or any other workspace data. It answers from the help articles and knows only your plan and role, so it can tell you whether a feature is on your plan.
  • What you can do: check a draft before you keep it. Auto Translate, Quick Setup, Quick Scene and the others hand you a suggestion; nothing is saved until you save it.
  • Good to know: help bot conversations are kept in Boardssey so the team can follow up and see which questions the articles miss. Dictation audio is not stored by Boardssey: only the text you keep in your notes is saved.

What the AI providers keep

The features above use Anthropic's and OpenAI's APIs, the business services, not the Claude or ChatGPT apps.

Helpful details

  • Key areas: neither company uses API data to train its models. Anthropic deletes requests and answers within 30 days. OpenAI may keep them for up to 30 days to detect abuse, then deletes them; audio sent for dictation is not kept. Either company may keep a request longer only if its safety systems flag it as breaking its usage policy.
  • What you can do: read the companies' own pages on how Anthropic stores API data and how OpenAI handles API data.
  • Good to know: the 30 days are the providers' safety window. When Anthropic's safety checks decline a Quick Setup, Quick Scene, Quick Insert or Quick Punchboard request, it is retried on another Claude model, still inside Anthropic. Boardssey keeps no log of what these features send; only the results you save stay in your workspace, like the rest of your data, plus the help bot conversations described above.

Claude and other AI assistants you connect

You can connect Claude, or another assistant, to Boardssey so it can read your games and, if you allow it, edit them. This only happens if you set it up yourself. See Connect Claude and other AI assistants.

Helpful details

  • Key areas: the assistant reads your data with your own access, only while you chat with it, and only the parts you allowed on the consent screen. Contacts, outreach and invoices stay hidden unless you tick Include contacts, outreach and invoices.
  • What you can do: disconnect it at any time under Settings > My Account > Apps with access to Boardssey.
  • Good to know: what the assistant reads becomes part of your conversation in that assistant, so the assistant's own terms and settings apply, not Boardssey's. With Claude, Anthropic does not train on conversations from its business plans (Team, Enterprise) or its API by default; on the Free, Pro and Max plans, it trains on your chats only if you allow it in Claude's privacy settings. Check your assistant's settings if this matters for a design.

Keep a design away from AI completely

If a game is confidential, you can keep it out of every AI feature.

Helpful details

  • Key areas: don't use the AI commands in its rules or notes, leave Use the rules of {game} unticked in Quick Setup, and don't connect an AI assistant (or connect one read only, and don't ask it about that game).
  • What you can do: share the game with a Collaborator instead of a full member when someone only needs that game. See Collaborator permissions strategy.
  • Good to know: everything else in Boardssey (storing, editing, playtesting, exporting, catalogs) works without any AI.

Tips & common questions

Does Boardssey use my game to train its own AI? No. Boardssey does not train AI models, and your data is never used to train anyone else's.

Can other Boardssey customers' AI answers include my game? No. Each request carries only your own text for that one answer, and the provider does not learn from it.

Is my artwork sent to an AI image generator? No. No feature sends your artwork, card images or 3D files to an AI model.

Can a teammate send my game to an AI? A teammate who can open a game can use these features on it, as they can copy its text anywhere. Give people only the access they need. See Roles explained.

Where can I ask more? Write to the team from the chat bubble or at support@boardssey.com. See Contacting the Boardssey team.

Still stuck? Write to support@boardssey.com or use the chat bubble.