Answer: Website-provided tool definitions and results are treated as untrusted content, and in the built-in browser each tool invocation receives a safety review before it runs, with normal website-access and confirmation policies still applying. Users can turn off site tools via Settings > Browser > Permissions. Status: Confirmed Entry: tool/chatgpt/permissions-and-security/webmcp-security-controls Revision: 1 Digest: sha256:1b395532cfdfc83e (changes whenever the content changes) Timeline: Created: 2026-11-13T03:54:54Z Last changed: 2026-11-13T03:54:54Z Last verified: 2026-11-20T13:43:33Z Next check: 2026-11-21T13:43:33Z Check interval: Daily Expected change: No fixed schedule Data: Field | Value | Source | As of Trust level of tool content | untrusted content | [1] | 2026-11-20 Safety review timing | each tool invocation receives a safety review before it runs | [1] | 2026-11-20 Setting to disable site tools | Enable site tools | [1] | 2026-11-20 Settings path | Settings > Browser > Permissions | [1] | 2026-11-20 Sources: [1] OpenAI. "Site tools". https://learn.chatgpt.com/docs/webmcp.md Role: primary. Accessed: 2026-11-20T13:43:33Z Quote: "Website-provided tool definitions and results are untrusted content." (Supports: Trust level of tool content) Quote: "each tool invocation receives a safety review before it runs" (Supports: Safety review timing) Quote: "You can turn off **Enable site tools** in **Settings > Browser > Permissions**." (Supports: Setting to disable site tools) Quote: "in **Settings > Browser > Permissions**" (Supports: Settings path) Snapshot SHA-256: 71a03cd7fbff2eaee11fac91ffc4b3df60b5b222841ba2aaad7f39a6a31397ce (captured 2026-11-13T03:54:53Z) Verification: Method: Automated extraction from [1] with deepseek-flash (JSON output, thinking disabled). Every value was checked by code to appear verbatim in a quoted passage of [1]. On this check the text of [1] was unchanged since 2026-11-13T03:54:54Z, so the values verified then were kept. Cross-check: not-available at 2026-11-20T13:43:33Z. No independent second source is configured for this entry Last successful check: 2026-11-20T13:43:33Z Last attempt: 2026-11-20T13:43:33Z Notes and limitations: Extracted automatically from a single official source (section "Security and user controls"). Related: https://latestn.com/tool/chatgpt/permissions-and-security/who-can-update-workspace-name Who can update the workspace name or image in ChatGPT Enterprise? https://latestn.com/tool/chatgpt/permissions-and-security/why-login-verification-required Why is ChatGPT asking me to verify my login? https://latestn.com/tool/chatgpt/permissions-and-security/workspace-analytics-permissions Who can view workspace analytics in ChatGPT Enterprise or Edu? https://latestn.com/tool/chatgpt/permissions-and-security/turn-mfa-on-off How do I turn multi-factor authentication (MFA) on or off for my ChatGPT account? https://latestn.com/tool/chatgpt/permissions-and-security/unexpected-approval-prompts Why does ChatGPT ask for approval unexpectedly when using an app? https://latestn.com/tool/chatgpt/permissions-and-security/verify-openai-social-media How can I verify official OpenAI social media accounts? https://latestn.com/tool/ All entries in AI tools Change log: Time | Event | Change | Source 2026-11-13T03:54:54Z | added | Entry created | — Changes for this entry as a feed: https://latestn.com/tool/chatgpt/permissions-and-security/webmcp-security-controls/latest Challenge this entry: Open challenges: None How: If a value here is wrong or out of date, send what is wrong and a link to evidence. Challenges are queued and checked against the evidence; they do not change the entry by themselves. If the entry changes as a result, the change log says so. Form: https://latestn.com/dispute/tool/chatgpt/permissions-and-security/webmcp-security-controls POST: https://latestn.com/dispute/tool/chatgpt/permissions-and-security/webmcp-security-controls with claim, evidence_url, and optionally field and contact (form-encoded or JSON) GET: https://latestn.com/dispute/tool/chatgpt/permissions-and-security/webmcp-security-controls?claim={what is wrong}&evidence_url={link} How to cite: LatestN. (2026). How do security and user controls work for WebMCP site tools in ChatGPT? Revision 1, last verified 2026-11-20T13:43:33Z. https://latestn.com/tool/chatgpt/permissions-and-security/webmcp-security-controls (accessed 2026-11-21T10:06:08Z). @misc{LatestN_tool_chatgpt_permissions_and_security_webmcp_security_controls_r1, title = {How do security and user controls work for WebMCP site tools in ChatGPT?}, author = {{LatestN}}, year = {2026}, howpublished = {\url{https://latestn.com/tool/chatgpt/permissions-and-security/webmcp-security-controls}}, note = {Revision 1; digest sha256:1b395532cfdfc83e; last verified 2026-11-20T13:43:33Z; accessed 2026-11-21T10:06:08Z} } Reuse: License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) Attribution: Copying and adapting are allowed, including commercially, if you credit LatestN and link to the entry you used. Each entry ends with a ready-made citation. URL: https://latestn.com/tool/chatgpt/permissions-and-security/webmcp-security-controls