The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Short answer: not with a documented, permanent news-site block in ordinary ChatGPT. You can tell ChatGPT to use primary or official sources for a particular answer and ask it to search again with a named source, date, or location. For a stronger per-task restriction, Deep Research lets you limit a task to domains you select. Enterprise and Edu administrators can disable web search altogether, but that removes search broadly rather than excluding news publishers alone.
These controls reduce or redirect news citations; none is documented as a universal guarantee for every ChatGPT response. Always inspect the Sources list and open cited pages before relying on an answer.
What “stop ChatGPT from using news websites” can mean
People usually mean one of two different things:
- Source exclusion: prevent news publishers from appearing in the citations attached to an answer.
- Source preference: make ChatGPT rely on primary documents, government pages, standards, academic papers, company documentation, or other sources you name instead of news coverage.
The distinction matters because ChatGPT has several controls with different scope. A prompt can steer one answer. Deep Research can constrain one research task. Workspace administrators can turn web search off for users. None of those is the same as a consumer-wide domain blocklist that permanently excludes every news website.
Recommended Free Tools
OpenAI’s web-search documentation does not describe a persistent setting for ordinary ChatGPT that blocks all news domains from search results or citations.
#1 Best Overall
Ordinary ChatGPT web search: steer each answer
In a normal ChatGPT conversation, the practical method is to state your source rule in the request, then check whether the returned citations follow it. OpenAI documents that you can ask ChatGPT to search again using a specific source, date, or location. That supports targeted retries, but it is not documented as a permanent exclusion preference.
Use an explicit source instruction
Put the restriction and the fallback behavior in the same prompt. For example:
Answer using only primary sources and official documentation. Do not use or cite news websites. If you cannot answer from those sources, say what remains unknown.
Recommended: Fix Windows Errors and Clear Junk Files in Minutes - Free Scan →Recommended: Update Every Outdated Driver on Your PC in One Scan - Free →Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.
“Primary sources” should be made concrete when the subject allows it. You might name a regulator, a court docket, a standards body, a company’s documentation, or a government statistical agency. The more precise the source class, the easier it is for ChatGPT to follow your intent.
Check the result instead of trusting the instruction
- Submit the question with the source restriction.
- Open the response’s Sources list or citation links.
- Check the domain and the page itself, not just the publisher name shown in the answer.
- Confirm that the cited page actually supports the sentence beside it.
- If a news site appears, ask ChatGPT to search again using a named official source or a narrower source type.
A useful retry is: “Search again using only the official documentation from [named organization]. Do not cite news sites. If that documentation does not establish the answer, say so.” OpenAI’s help page supports asking for a new search with a specific source, but it does not promise that every requested source will be available or that a retry will always produce a citation-free result.
Use dates and locations when they matter
News coverage is often returned because the question is time-sensitive or geographically ambiguous. Add a date range, jurisdiction, or location when appropriate: “Use official filings published in 2025,” or “Use the regulator’s pages for the United Kingdom.” OpenAI specifically lists source, date, and location as criteria users can provide when asking ChatGPT to search again. These constraints improve relevance; they do not create a permanent domain filter.
Tell ChatGPT what to do when the evidence is insufficient
Do not ask it to fill gaps from news coverage after you have prohibited those sources. Include a stopping rule such as “If no qualifying source supports the claim, state that the evidence is insufficient.” This produces a more useful uncertainty statement than an answer that silently relaxes your source requirement.
Deep Research: the strongest per-task source control
Deep Research has a control that ordinary web search does not document: you can enter websites or domains and either restrict the task to those sites or prioritize them while still allowing broader web search. OpenAI describes this in its Deep Research FAQ.
Restrict versus prioritize
| Deep Research choice | What it does | When to use it | Does it exclude news domains? |
|---|---|---|---|
| Restrict to selected sites | Limits the research task to the domains you enter. | Use when every cited source must come from an approved list. | It is the stronger option for avoiding news sites, provided you do not include them in the selected domains. |
| Prioritize selected sites | Favors the domains you enter but still permits broader web search. | Use when official sources should lead but additional context may be useful. | Not reliably; news pages can still be used if the task is allowed to search the wider web. |
To avoid news sources in a Deep Research task, open the source controls, enter only the domains you approve, and choose the restriction mode rather than merely prioritizing those domains. When the report is complete, review its source list. The control applies to that task; it is not described as a permanent setting for all future chats.
Build a useful approved-domain list
A restriction works only if the selected sites contain enough material to answer the question. Include the authoritative domains that normally publish the evidence you need: an agency’s regulations and notices, a company’s documentation and filings, a standards organization, or a court and legislative database. If the list is too narrow, Deep Research may return an incomplete answer or report that it cannot establish a point. That is preferable to silently substituting disallowed news coverage, but you should plan the source list before starting a long task.
Verify the completed report
Source restrictions change where Deep Research looks; they do not remove the need for checking citations. Open the cited pages and make sure the publication date, jurisdiction, edition, and wording match the claim. OpenAI warns that search results and citations can be incomplete, outdated, or incorrect. A domain restriction is therefore a boundary on the search set, not a guarantee that every resulting statement is accurate.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Enterprise and Edu: administrators can turn search off
In Enterprise and Edu workspaces, owners and administrators can enable or disable ChatGPT search for affected users and for GPTs created in the workspace. The Enterprise and Edu documentation also describes role-based search permissions in some workspaces.
| Administrative action | Result for users | News-only filtering? |
|---|---|---|
| Disable ChatGPT search | Web search is unavailable to the users or workspace scope covered by the setting. | No. This is a broad web-search switch. |
| Enable search with role-based permissions | Access can be limited to permitted roles where the workspace supports that control. | No. It governs who can search, not which publisher domains appear. |
| Configure search for workspace GPTs | Administrators can control whether search is available to GPTs created in the workspace. | No. Availability is controlled; a news-domain blocklist is not documented. |
If an administrator disables search, ChatGPT can still answer from its built-in knowledge and from information supplied in the conversation, but the user cannot use the web-search feature. This may meet a strict no-web requirement, yet it also removes access to official websites and other current sources. It should therefore be treated as a workspace policy decision, not as a news-source preference.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.ChatGPT Atlas has a different setting
ChatGPT Atlas includes a “ChatGPT page visibility” setting. OpenAI’s Atlas web-browsing documentation describes that setting as controlling whether Atlas can read a site’s page content for on-page assistance. It is separate from chat history and Data controls.
Page visibility is not documented as a filter for ChatGPT Search citations. Turning it off does not establish that news domains will be excluded from search answers, and turning it on does not grant permission to cite a particular publisher. Treat Atlas page reading and ChatGPT Search as separate features.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Which option fits your goal?
| Goal | Best available control | Search remains available? | Scope and limitation |
|---|---|---|---|
| Keep one answer focused on official sources | Prompt for primary or named sources; retry with a specific source, date, or location. | Yes. | Per answer. The prompt is guidance, not a documented permanent exclusion. |
| Conduct a controlled multi-source investigation | Deep Research restricted to selected domains. | Yes, within the approved domains. | Per task. The selected sites must contain enough evidence. |
| Prefer official sources but allow wider context | Deep Research with selected domains prioritized. | Yes, across the wider web. | News websites may still appear. |
| Prevent users from browsing the web in a managed workspace | Enterprise or Edu administrator disables ChatGPT search. | No for affected users or scopes. | Broad switch; it does not target news sites alone. |
| Stop Atlas from reading pages for on-page help | Adjust ChatGPT page visibility in Atlas. | Unchanged as a Search question. | Page-reading control, not a citation filter. |
Why a news citation can still appear
Several failure modes are easy to misread as a missing setting:
- A prompt is not a persistent preference. A request applies to the answer or task in which you make it. Starting a new conversation without the instruction does not carry the restriction forward as a documented account-wide rule.
- Prioritizing is not restricting. In Deep Research, prioritization still permits full-web search. Choose restriction when disallowed domains must be outside the task’s source set.
- The request may be too broad. If you ask for “the latest reaction” but allow only a regulator’s site, the approved source may not contain that information. Narrow the claim, expand the approved primary-source list, or accept an explicit “not established” result rather than relaxing the rule silently.
- Citations need independent checking. OpenAI states that search results and citations can be incomplete, outdated, or incorrect. A citation that looks authoritative may not support the exact wording of the answer.
A repeatable no-news workflow
- Define the evidence standard. Decide whether you need primary documents only, official documentation, a named institution, or a specific jurisdiction and date.
- Choose the control. Use a prompt for one answer, Deep Research restriction for a bounded task, or an administrator’s search policy for a managed workspace.
- State the fallback. Require ChatGPT to identify what cannot be established from the approved sources instead of filling the gap with news coverage.
- Review every citation. Check the domain, page title, publication date, and the passage supporting the claim.
- Retry narrowly if needed. Ask for a search using a named official source, date, or location, as OpenAI’s web-search help recommends.
- Record unresolved points. If the approved sources disagree or contain no answer, preserve that uncertainty; do not treat a news article that appears later as proof that the original restriction worked.
Bottom line
There is no documented, permanent consumer setting in ordinary ChatGPT that blocks every news website from being used or cited. Prompt steering is useful for individual answers, Deep Research’s restricted-domain mode is the strongest per-task option, and Enterprise or Edu administrators can disable web search entirely. Atlas page visibility addresses page reading, not search citations. Whatever method you use, inspect the source list and verify that each citation supports the claim.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
