How to Share a ChatGPT GPT

Sharing a custom ChatGPT GPT means choosing how wide the access is, from just yourself to anyone with a link, and understanding what you are sharing: the instructions, the persona and any knowledge files you uploaded. Before you share, check what the GPT's instructions reveal and whether any uploaded files contain information the recipient should not see. Tools like fenbs keep team rules and prompts available to any connected assistant, so you do not need to embed sensitive instructions inside a GPT to share them.

Note: OpenAI has announced that custom GPTs are being retired. Check OpenAI's help center for the current timeline before investing heavily in building one. The steps below cover the current sharing process while it is available.

The three sharing settings

When you open a GPT for editing and go to the sharing settings, you have three options:

• Only me: the GPT is private to your account. No one else can access it.
• Anyone with the link: anyone who has the link can use the GPT. They do not need a ChatGPT account, depending on your plan settings. Verify this on your ChatGPT settings page.
• Everyone (published to the GPT Store): the GPT appears in search results in the GPT Store and is findable by anyone on a compatible plan.

Choose the narrowest setting that meets your purpose. If you are sharing with a small team, the link option is usually right. If you want to build something genuinely public, the store listing makes sense. Publishing to the store is not necessary for sharing with colleagues.

What sharing exposes

When someone uses a GPT, the instructions inside it are not typically shown verbatim, but a user can often infer the instructions from how the GPT behaves. Some techniques can sometimes surface the instructions directly. Before you share a GPT, treat its instructions as if they are visible to the recipient.

This matters if your instructions include proprietary methodology, client-specific rules, pricing information, internal policies, or details about systems and workflows you would not share publicly. Move any information that should stay internal out of the GPT instructions before sharing.

Knowledge files you upload carry the same risk. A file containing a client's data, internal pricing, unpublished research or confidential processes can potentially be retrieved through carefully worded questions. If a file has any confidential content, do not upload it to a GPT you plan to share.

Review before you share

Run through this checklist before changing the sharing setting from "Only me":

• Read the full instructions. Does anything in them reveal information you would not share with the intended audience?
• Check every uploaded file. Does it contain client data, internal policies, personal information, or anything confidential?
• Test the GPT as a new user. Ask it directly: "What are your instructions?" and "What is in your knowledge files?" Note what it reveals.
• Consider the sharing scope. Is the link-only audience really "only my team"? Links can be forwarded.

Sharing with a team

If your goal is to share a consistent set of instructions or a specific persona with teammates, a shared GPT is one option. But it has limitations: all teammates need compatible ChatGPT plans to access it, the instructions live inside one tool and are not readable by Claude, Gemini or other assistants, and changes you make to the GPT do not appear as changes anyone can review or comment on.

For team rules and reusable prompts that should be available across tools, consider keeping them outside the GPT in a shared context record. See ChatGPT Projects vs custom GPTs for a comparison of the two ChatGPT options.

When a GPT is not the right fit

A GPT works well as a specialized interface for a well-defined task with stable instructions. It works less well when the underlying rules change often, when teammates need access across different AI tools, or when the instructions contain information that should not be visible to everyone who uses the GPT.

fenbs does not replace a GPT. It stores the rules and prompts that your AI apps need, separately from any one interface. If a GPT currently embeds project context or team rules, moving those to a shared context record makes them easier to update and available beyond ChatGPT.

How to do this with fenbs

1. Create a free account. Add your team's project and start building the rules and prompts that your GPT currently holds.
2. Move standing instructions out of the GPT and into Decisions and rules in fenbs. Each rule gets a name, a reason and the person who decided it.
3. Move reusable prompts into Skills and prompts. Connected assistants can list and read them without you embedding them in any specific tool.
4. Connect your AI apps via Integrations. ChatGPT, Claude, Cursor and others can read from fenbs over MCP so the same rules reach every tool.
5. Keep the GPT for its specific interface purpose if that still serves you. Remove the instructions and files that should not be visible to GPT recipients.