Reading fact-check flags

Every post Blog AI writes is fact-checked before you see it. The writer cites a source for each fact it uses; then a separate check, run by a different AI model from the one that wrote the post, opens each cited source and compares it with what the post says. The results are in the Fact check panel on the post's page.

One rule is enforced by the system itself, not just the screen: a statistic that doesn't match its source can't be published until you resolve it. This article explains the verdicts, how to resolve a flag, and what the check can't do.

What gets checked

• Every cited claim. Each fact the writer used is recorded with its source link and the passage it relied on. The check fetches the source and decides whether it supports the claim.
• Numbers without a source. The check also reads the whole post for statistics that no cited claim covers, such as a percentage that appears with no link.

Refreshes and merges are checked the same way, but only on the text that changed.

Facts about your own business come from you, not a web page. Figures from your answers to Add what you know show as From you, and figures from your site's Facts about your business as From your business facts. Neither blocks publishing, unless the post changed the figure: then it shows as Not in your note or Not in your business facts and must be fixed.

The verdicts

The panel's heading shows how many items need fixing, or Clear to publish when none do. Checked facts that passed are listed underneath, each linked to its source, so you can spot-check them too.

Resolve a flag

1. Open the Fact check panel. On the post's page, open Fact check in the side panel. A post that needs review also shows Needs review next to the publishing buttons, and appears on the Actions page as Review claims.
2. Read the flagged claim and its source. Each flag shows the claim, the verdict, what the source says (Source says:) and a link to the source. Open the source and look for yourself.
3. Choose one of the actions on the flag.
   • Remove this sentence. Blog AI finds the sentence in the post (the body, a key takeaway or an FAQ answer) and shows you the paragraph before and after. Click Remove it to make the change; Blog AI saves the post and records a version you can go back to. If the sentence is the whole paragraph, list item, table row or FAQ answer, that goes too.
   • Rewrite it. Type a new version of the sentence without the figure, or click Suggest a rewrite and Blog AI drafts one for you to edit (free). Click Preview change, then Save rewrite. If you'd rather change the post yourself in the editor (for example to use a figure from a source you trust), do that, then click tell Blog AI you've rewritten it.
   • I've checked the source (cited claims only). You opened the source and it does support the claim.
   • The figure is right — remove the broken link (cited claims only). See When the source page no longer works.
4. Check the panel. It updates, and when nothing blocking is left it says Clear to publish.
5. Publish. Publish now and Schedule for work again.

Removing and rewriting are checked. A flag only counts as removed or rewritten once its text is no longer in the post, including the key takeaways and FAQ. If it's still there, Blog AI says so and leaves the flag open, and publishing stays off.

Sometimes Blog AI can't change the sentence for you: it was reworded since the check, it appears in more than one place with different wording, it's in the title, an FAQ question or a table's header row, or the same figure is also somewhere else in the post. Blog AI doesn't guess in those cases. It tells you why, and you edit the post in the editor instead.

"I've checked the source" is for cases where the check got it wrong, which happens: a figure buried in a PDF or a table, or phrased differently from the post. Only use it after you've read the source yourself. You're publishing under your name, and the fact check records that you confirmed it.

When the source page no longer works

When a claim's source page has gone, the flag says Source page no longer works and offers The figure is right — remove the broken link first. Use it when you know the number is right.

1. Not sure it's gone? Check the link again tries the page once more.
2. Click The figure is right — remove the broken link. You see the paragraph before and after. Only the link to that page goes: its words and the number stay. Links to other pages are left alone.
3. Or paste a page that works into …or replace it with a working link and click Preview with this link. The link keeps its words and goes to the new page.
4. Click Remove the link (or Replace the link). Blog AI saves the post and notes that you checked the figure yourself.

If Blog AI can't find the link (say it's in a list of sources), change it in the editor, then click tell Blog AI you've removed it. Blog AI checks it's gone.

Why unsupported statistics block publishing

Numbers are what readers remember and repeat, and they're where AI writing most often goes wrong. A made-up percentage under your name costs you trust that no amount of traffic wins back. So Blog AI treats this as a hard rule: the publish buttons stay off, and a check in the database refuses the publish even if something tried to go around the screen.

Other flags, such as a non-numerical claim that doesn't match, are shown for you to judge but don't block publishing.

What the check can't do

The fact check makes errors less likely. It doesn't make a post correct by itself.

• It checks claims against their cited sources, not against the whole world. If a source is wrong, a claim that matches it still passes.
• It's an AI model, so it can be wrong both ways. It can flag a claim that's fine (you clear it with I've checked the source) and, rarely, pass one that isn't quite right.
• It can't read everything. Pages behind logins, paywalls or bot blocking come back as Couldn't open source.
• It doesn't judge advice or opinion. "You should update old posts" isn't a fact to check. Whether advice suits your business is your call.
• It doesn't check posts you write by hand, or text you add after the check ran.

The terms of service say the same thing in legal language: you're responsible for what you publish. Read every post before it goes out.

Common questions

A flag says "Still in the post"

The flag was marked as removed or rewritten, but its text is still in the post (before October 2026 the buttons only recorded what you said you'd done). Use Remove this sentence or Rewrite it on it. Until then the post can't be published, scheduled or updated live. If the post is already live, use Update live post afterwards so your site gets the change.

I removed a sentence from a live post. Is it gone from my site?

Not yet. Blog AI changes its own copy of the post. Click Update live post to send the change to your site.

A restored version brought flags back

Restoring an earlier version brings back the claims that version contained. Any you'd resolved are flagged again so you can check them against the restored text.

Why is a claim flagged when the source clearly says it?

Usually because the wording or the number differs slightly (rounded, a different year, a different country), or the source was slow to load. Read it, and if it's right, click I've checked the source.

Can I turn the fact check off?

No. It's part of how every post is written, and it costs you nothing extra: the check is included in the price of a post.