Why Claude Failed to Fetch Links: Causes and Fixes

When Claude says it failed to fetch a link, the reason is almost always one of a small set of causes: the page requires a login or payment, the server blocks AI tools, the file format is not supported, or the URL has changed or been removed. Most of these have a workaround that does not involve bypassing any restriction.

The steps below cover the most common situations. If fetching consistently fails across multiple links, check whether Claude is experiencing an outage before spending time on individual URLs. A tool like fenbs can keep a record of the sources you have already confirmed as accessible, along with the findings you accepted from each one.

Steps to try first

1. Open the URL yourself in a browser and confirm the page loads without logging in. If it does not, see the section below on login and paywall pages.
2. Check that the URL is complete and correct. A missing character or a copied line break can produce a 404 that looks like a fetch failure.
3. Try the URL again in a new message. Transient network errors are common and often resolve on a second attempt.
4. If the page loads in your browser but Claude still cannot read it, copy and paste the key text from the page directly into the chat instead.
5. For a PDF or video file, copy the text or transcript yourself and paste it in. Claude cannot play video or audio, and some PDF hosting services block AI fetching.

Pages that require a login or payment

A login wall or paywall is the most common cause of a failed fetch. Claude cannot log in to a site, accept cookies on your behalf or use a session you have already opened in your browser. If you can read the page only because you are subscribed or signed in, Claude will see a login prompt or a partial preview instead.

The workaround is to copy the content you need from the page and paste it into the chat directly. If the information is behind a paywall you subscribe to, you can legally read it and share the relevant excerpt for analysis. Do not paste an entire article where that would breach the site's terms of service.

File types Claude cannot open from a URL

Claude can read HTML pages and some plain-text formats when fetching from a URL. It cannot open video files, audio files, password-protected PDFs, or formats that require a proprietary reader. Some PDF hosting services also block fetching by returning a download prompt instead of the document itself.

For a PDF you can open, download it, then upload the file directly to the conversation. For a video, find or create a transcript and paste that in. For a spreadsheet or slide deck, export the relevant content as text.

Pages that block AI tools

Some sites use their robots.txt file or server-side rules to block AI crawlers. When Claude encounters one of these, it will typically say the fetch failed or that it could not access the content. This is the site's deliberate choice, not a bug in Claude.

Do not try to circumvent these restrictions. If the site has blocked automated access, find an alternative primary source for the same information: an official press release, a regulatory filing, a vendor's own documentation page, or a cached version the site itself has made available.

Moved and removed pages

URLs that worked last month may return a 404 or a redirect that Claude cannot follow. This is common with news articles, documentation pages and pages that have been reorganized. A URL that was stable for years can change after a site redesign or a domain migration.

Search for the page title to find its new location. For regulatory or legal documents, look for the primary agency or court's own search function rather than relying on a URL that may change. If the original page is genuinely gone and you need the content, check whether an archive copy is available through a web archiving service.

Keeping track of sources that work

When you are building a piece of work that draws on multiple sources, tracking which URLs are accessible and what each one confirmed is valuable. Losing a working source link mid-project means going back to find it again, and forgetting which claims were checked against which sources leads to errors down the line.

How to do this with fenbs

1. Create a free account and open the project you are researching.
2. Each time you confirm a source is accessible and reliable, add its URL to Links and listings with a short note on what it supports.
3. Record the specific claims you accepted as context notes in the project, linked to the source.
4. Connect your AI assistant over MCP from Integrations and ask it to read the project context before continuing the research.
5. When a source moves or goes offline, update its entry with the new URL or mark it as no longer available so a future session does not waste time on a dead link.