Ask an AI assistant about "the best project management tools" or "current mortgage rules" and you want an answer that reflects today, not three years ago. AI search systems know this, and for recency-sensitive questions they appear to lean towards content that is recent or recently updated. That has created a tempting shortcut: change the date at the top of the article and hope the machines notice.
They might notice, but not in the way you hope. Google explicitly lists fake date changes as a warning sign of search-engine-first content, and Bing says it uses sitemap lastmod values only when they reflect real changes. The freshness signals that work are the honest ones: visible dates, accurate lastmod values, and updates that genuinely change the content.
This guide explains how dates, sitemaps and real updates influence recency-sensitive AI answers, what the evidence says, and how to build an update process that earns freshness rather than faking it.
Key Takeaways
- An Ahrefs study (July 2025, about 17 million citations) found AI assistants cited somewhat fresher content than organic search on average, with Google AI Overviews the exception.
- Google uses sitemap lastmod only when it is consistently and verifiably accurate; Bing treats an accurate lastmod as a key freshness signal for crawling.
- Show a clear "Published" and "Last updated" date and keep it consistent with datePublished and dateModified in structured data.
- Changing dates without substantial content changes is something Google explicitly warns against.
- Freshness is one factor among many: older, authoritative content still gets cited a lot.
What the evidence says about freshness and AI answers
The largest public analysis is Ahrefs’ "Do AI assistants prefer to cite fresh content?", published on 28 July 2025. It compared roughly 17 million URLs cited by ChatGPT, Perplexity, Gemini, Copilot and Google AI Overviews with organic search results. Its findings:
- Cited URLs averaged 1,064 days since publication versus 1,432 days for organic results, about 25.7% "fresher".
- On last-updated dates the gap was smaller: 909 versus 1,047 days, about 13.1%.
- ChatGPT showed the strongest preference for newer content; Google AI Overviews cited slightly older content than organic results.
- The average cited page was still about 2.9 years old, so long-lived content continues to be cited.
These are one study’s correlations, measured with Ahrefs’ own crawl dates. They do not prove that updating a page causes citations. They do suggest that, for questions where recency matters, being genuinely current is an advantage, especially in ChatGPT and Perplexity.
Infrastructure providers are paying attention too. In July 2026, Cloudflare announced a research programme to make AI search smarter using network signals such as content freshness, partly so answer engines can avoid recrawling unchanged material. And on the crawler side, OpenAI’s crawler documentation notes that robots.txt changes for search can take about 24 hours to take effect, a reminder that AI systems discover changes through crawling, not instantly.
The three freshness signals you control
| Signal | What search engines say | How to get it right |
|---|---|---|
| Visible dates | Google recommends a user-visible date, featured prominently, labelled e.g. "Published" or "Last updated" | Show both dates near the headline; avoid other confusing dates |
| Structured data dates | Google recommends datePublished and/or dateModified in Article-type markup, consistent with visible dates | Generate both from the CMS, never hard-code |
| Sitemap lastmod | Google uses it only if consistently and verifiably accurate; Bing uses it to decide what to recrawl | Update only on significant changes; use full date and time |
| Real content changes | Google lists main content, structured data or links as significant updates; a copyright date is not | Change facts, data, sections, examples, not just wording |
| Change notification | Bing recommends IndexNow alongside sitemaps for faster discovery | Ping IndexNow on publish and significant updates |
Sources: Google’s byline date guidance, Google’s sitemap documentation and Bing’s July 2025 sitemap post. For the mechanics of sitemaps and IndexNow, see our guides to XML sitemaps and IndexNow; for markup, see Article schema.
Why fake date-bumping backfires
Google’s guide to creating helpful, reliable content includes a direct self-assessment question about this exact tactic.
“Are you changing the date of pages to make them seem fresh when the content has not substantially changed?”
— Google Search Central, Creating helpful, reliable, people-first content
Bing is equally clear about sitemaps, advising site owners to avoid setting lastmod to the time the sitemap was generated unless the URL’s content actually changed. There are three practical risks:
- Lost trust in your lastmod. If your dates change without real changes, search engines can stop relying on them, which slows discovery of genuine updates.
- Wasted crawling. Crawlers revisit pages that have not changed, which adds load (see our new guide on AI crawler bandwidth).
- Reader trust. A "Updated October 2026" label on an article that cites 2023 prices is obvious to readers and can end up quoted in AI answers as outdated.
What counts as a real update?
| Real update (change dateModified) | Not a real update (keep the date) |
|---|---|
| New data, prices, statistics or regulations | Fixing a typo or changing a word |
| Adding a new section that answers a new question | Changing the featured image |
| Replacing outdated screenshots and steps | Updating the copyright year in the footer |
| Rewriting recommendations after testing | Re-ordering paragraphs |
| Removing incorrect or obsolete advice | Adding internal links only |
When in doubt, ask whether a returning reader would notice and benefit from the change. Our existing guide on how to update old content covers the editorial side in depth.
Step-by-step: a freshness process for AI search
- Identify recency-sensitive pages. Pricing, comparisons, "best" lists, statistics, regulations, product specs and yearly guides.
- Set review cycles by volatility: monthly for prices and rankings, quarterly for comparisons, yearly for evergreen guides.
- Update substance first. Refresh facts, data and examples; cite dated sources (our new guide to statistics pages for AI citations shows how).
- Then update dates. Change the visible "Last updated", dateModified and sitemap lastmod together, automatically from the CMS.
- Notify. Submit the updated URL via IndexNow and keep the sitemap current.
- Check pickup. Look for AI crawler and user-triggered fetches of the URL in your logs (see AI bots in server logs).
- Measure citations before and after, accepting that AI answers vary from run to run (our new guide to AI answer variability explains why).
Common mistakes
- Automatic "today’s date" stamps. Some themes show the current date on every page view; remove them.
- lastmod equals sitemap build time. Bing specifically warns against it.
- Mismatched dates. Visible date, dateModified and lastmod should agree.
- Putting the year in titles without updating content. "Best X in 2026" with 2024 data is a credibility problem.
- Republishing under new URLs. You lose links and history; update in place instead.
- Updating everything at once. Prioritise pages where recency changes the answer.
Related Guides
- Syndicated and Duplicate Content in AI Answers: Who Gets the Citation?
- Paywalls and AI Search: Visibility Without Giving Content Away
- International AI Search: How Language and Country Shape AI Answers
Frequently Asked Questions
Do AI assistants prefer fresh content?
An Ahrefs study of about 17 million citations (July 2025) found AI assistants cited somewhat newer content than organic results on average, with ChatGPT showing the strongest preference and Google AI Overviews the weakest.
Does Google use the lastmod date in sitemaps?
Google says it uses lastmod only if it is consistently and verifiably accurate, reflecting the last significant update to the page.
Can I just change the date on my article?
Google explicitly warns against changing dates to make pages seem fresh when the content has not substantially changed. Update the content first.
Should I show both published and updated dates?
Yes. Google recommends a prominent user-visible date and consistent datePublished and dateModified values in structured data.
How often should I update content for AI search?
It depends on volatility. Pricing and comparison pages may need monthly reviews; evergreen guides may only need a yearly check.
Conclusion
Freshness in AI search is earned, not declared. Keep your visible dates, structured data and sitemap lastmod honest and in sync, update the pages where recency really matters, and let the signals follow the substance. Our content writing team can run a structured refresh programme for your key pages, or you can request a custom quote.
References
- Ahrefs: Do AI assistants prefer to cite fresh content? (Jul 2025)
- Google Search Central: Build and submit a sitemap
- Google Search Central: Publication (byline) dates guidance
- Google Search Central: Creating helpful, reliable, people-first content
- Bing Webmaster Blog: Keeping content discoverable with sitemaps in AI-powered search (Jul 2025)
- Cloudflare: Making AI search smarter (Jul 2026)
- OpenAI: Overview of OpenAI crawlers
