How long does it take AI to update what it says about my brand?

There is no single update time, because AI answers come from two different places.

• Retrieval-on answers use live search. The engine runs queries, pulls pages from its index, and writes from them. These answers can change once your updated page is recrawled and reindexed.
• Retrieval-off answers come from what the model learned in training. That knowledge is frozen at a cutoff date. Nothing you publish today reaches it until the vendor trains and ships a new model.

No vendor publishes how often it recrawls a given site, and none publishes a release schedule for new models. So the honest answer is: search-backed answers update on the crawler's schedule, and trained answers update on the vendor's release schedule. You control neither clock directly.

What you do control is whether anything blocks the first clock. That is where most fixable delays live. For a platform-level walkthrough, see how trained knowledge and search differ in ChatGPT.

A search-backed answer can only reflect your new facts after four things are true.

1. The right crawler is allowed. robots.txt rules are per bot. OpenAI's crawler documentation says OAI-SearchBot controls ChatGPT search, while GPTBot governs training only, and that each setting is independent. Blocking the search bot keeps you out of ChatGPT search answers.
2. The crawler can actually reach the page. A firewall or CDN can turn away a crawler that robots.txt allows. OpenAI's help center says eligibility needs both the robots.txt permission and a host that allows its published search bot IP addresses.
3. The new fact is in the HTML. Vercel found that ChatGPT and Claude crawlers fetch JavaScript files but don't execute them. A price that only appears after JavaScript runs may be invisible to them. Googlebot is the exception, because it renders.
4. The page is in that engine's index. Google says a page must be indexed and eligible to show with a snippet to appear in AI Overviews or AI Mode, and that indexing and serving aren't guaranteed.

Each engine reads its own index. In Otterly's 2026 experiment, Google deindexed two test sites and its AI surfaces almost stopped citing them, while Microsoft Copilot, which grounds on Bing, kept citing them. An update can show up in one engine and not another for the same reason.

When an engine answers without searching, it relies on training data. That data has a hard edge.

As of October 2026, OpenAI's model documentation lists a February 16, 2026 knowledge cutoff for its GPT-5.6 models, and an August 31, 2025 cutoff for the faster Instant path that ChatGPT also routes to. That means "what ChatGPT knows" is not one date. A rebrand, price change or acquisition after those dates is unknown to those models whenever they answer without search.

No publisher action reaches a trained answer this quarter. Allowing training crawlers like GPTBot is the only theoretical route into a future model, but nobody has measured whether it changes what a model says, and any effect would arrive a model generation later.

The practical move is to know which of your changes fall after each cutoff. The free knowledge cutoff check compares your brand news against model cutoffs, and the playbook on correcting outdated facts past a knowledge cutoff covers what to do about each one.

Even after your page is indexed, you may still see the old answer. Three things cause this.

Search is decided per question

Retrieval is discretionary on every platform. ChatGPT decides for itself when to search. On Claude.ai, web search is a user toggle. Google AI Overviews often don't trigger at all. So one question gets a current, search-backed answer and the next gets a stale trained one.

Other sites still say the old thing

Engines often cite third-party pages: directories, reviews, comparison posts, press coverage. Updating your site does not update theirs. An old article can keep feeding the old fact into search-backed answers. See whether old news stories shape AI answers today.

The error isn't outdated at all

Some wrong answers are fabrications, not stale facts. Waiting won't fix those. Telling a hallucination from an outdated answer decides which fix applies.

You can't set the timeline, but you can remove what blocks it.

1. Check crawler access per bot. Confirm robots.txt allows OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot. Then confirm your firewall doesn't return a 403 to them. The AI Crawler Checker requests a page as each AI crawler and shows which ones get turned away.
2. Put the new fact in plain HTML. State the specific number or name in one clear sentence that loads before JavaScript runs. That gives the engine a sentence it can quote.
3. Fix contradictions on your own site. If the pricing page, footer and about page disagree, engines have conflicting text to pick from.
4. Check each index separately. Use Google Search Console and Bing Webmaster Tools. One healthy index says nothing about the other.
5. Ask the sites that cite you to update. Being named correctly on those pages is associated with appearing correctly in answers.
6. Recheck with search on and off. The free checker on this site shows what engines say for up to three buyer questions, so you can see which answers have moved.

For pricing specifically, see why ChatGPT keeps getting pricing wrong.