Perplexity's new freshness filter changes who gets cited in AI search
Perplexity added last_updated_filter to its Search API in August 2026, turning content freshness into a hard citation gate. Pages without valid dateModified schema are excluded from filtered queries entirely.
B
Stephan Ochse
10 min read
August 26, 2026
Share
BrandCited monitors content freshness signals across nine AI search engines: ChatGPT, Claude, Gemini, Perplexity, Grok, DeepSeek, Copilot, Brave, and You.com. On August 22, 2026, Perplexity added a last_updated_filter to its Search API, turning content freshness from a soft ranking preference into a hard retrieval gate. Pages without a valid dateModified timestamp in Article schema — or pages that haven't been substantively updated — will be excluded from filtered queries entirely. Content refreshed within 30 days earns 3.2x more AI citations than comparable pages over 90 days old, according to Authority Tech's 2026 freshness analysis.
Run a free scan at brandcited.ai to see whether your content freshness signals pass or fail across Perplexity, ChatGPT, and seven other engines.
Perplexity's last_updated_filter is a Search API parameter that restricts results to content updated within a specified time window, excluding any result whose last-modified timestamp falls outside that window regardless of how well the content matches the query.
Atomic fact 1: Perplexity's August 2026 changelog lists last_updated_filter as a new parameter that enables filtering search results by when content was last updated, in addition to the existing published_after and published_before date fields. (Perplexity API changelog)
Atomic fact 2: Before this release, Perplexity already weighted freshness heavily — citation rates for content updated within 30 days reached 82%, dropping to 37% for content older than 180 days, according to . The new filter converts that preference into a binary gate for any query where the filter is applied.
Track your AI visibility now
See exactly how AI engines like ChatGPT, Perplexity, and Gemini perceive your brand.
The BrandCited team covers GEO, AI search optimization, and brand visibility strategy. We publish research, practical guides, and product updates every week.
The practical consequence is a complete exclusion, not a ranking penalty. If a developer builds an application or agent on Perplexity's API and applies last_updated_filter for recency-sensitive queries, brands with stale content fall out of the citation candidate pool entirely. The filter doesn't lower their citation probability. It sets it to zero.
This matters because Perplexity's API powers a growing number of enterprise applications — including, as of August 25, 2026, its new Computer agent integrated into Microsoft 365 apps. Every application built on that API can now apply the freshness gate, extending it beyond the Perplexity web interface.
How does content freshness affect citation rates across AI engines?#
Content freshness affects citation rates differently across the nine major AI engines, with Perplexity the most aggressive on recency and ChatGPT the most forgiving toward older authoritative content.
Atomic fact 1: Content updated within 30 days earns 3.2x more AI citations than content older than 90 days. A single substantive content update increased Perplexity citation frequency by 37% within the first 48 hours in controlled testing, according to Authority Tech's 2026 freshness study.
Atomic fact 2: Pages not refreshed within 90 days are three times more likely to drop out of AI citation pools than regularly updated content, according to Onely's 2026 brand AI visibility research.
Perplexity uses real-time web retrieval on every query. Its citation logic weighs publication date, last-modified date, and references to current data inside the page body. If your page cites 2022 statistics without an update note, Perplexity treats the entire page as 2022 content. ChatGPT and Claude rely more on training data, which makes them more tolerant of older authoritative content — but ChatGPT Search processes 250 to 500 million weekly queries in real-time mode, and that proportion is growing.
Content freshness matters most on three query types: comparison queries ("best X for Y right now"), news-adjacent queries where users expect current information, and queries where developers apply Perplexity's last_updated_filter or equivalent recency gates.
What breaks for brands without dateModified schema?#
Without a valid dateModified field in Article or WebPage schema, AI engines cannot determine when content was last updated — so they default to the original publication date, the crawl date, or nothing, all of which may be months older than the actual content.
Atomic fact 1: AI engines use the dateModified field in Article schema to determine content freshness. When the field is absent, systems default to the original publication date or an estimated crawl date, both of which may be significantly older than the actual content, according to Contently's 2026 schema implementation guide.
Atomic fact 2: The dateModified field ranks as the second-highest-impact individual schema property for AI citation rates — after the author field — in Derivatex's analysis of 73 websites, which found sites with complete schema markup earn 2.5x to 3.2x higher AI citation rates than comparable sites without it.
Three failure modes appear most often:
Cosmetic freshness: Changing only the dateModified timestamp without substantive content changes reduces retrieval trust. AI crawlers compare page snapshots over time and discount date changes that don't match real content changes.
Missing schema: Publishing a content update without a dateModified field makes the update invisible to AI retrieval systems.
Incomplete schema: Using datePublished only — without dateModified — signals a page that has never been updated, even if it has been refreshed multiple times.
A brand that refreshes its pricing page quarterly but skips updating the schema dateModified field appears to Perplexity as a page last touched at original publication. That page drops in citation priority and fails freshness filters entirely.
How do you update content to pass Perplexity's freshness filter?#
Two changes fix the majority of freshness failures: adding a valid dateModified field to Article schema and pairing that field with substantive content updates that correspond to the new date.
Atomic fact 1: Article schema with dateModified tells Perplexity, ChatGPT Browse, and Google AI Overviews exactly when content was last verified. Pages with this field show higher citation discovery rates in retrieval-augmented AI systems than pages relying on crawl-date inference, according to StackMatix's 2026 structured data guide.
Track your AI visibility for free
See how ChatGPT, Claude, Gemini, and 4 other AI platforms mention your brand.
Atomic fact 2: The dateModified field ranks as the second-highest-impact individual schema property for AI citation rates, after the author field, in Derivatex's analysis of 73 websites — which found sites with complete schema markup earn 2.5x to 3.2x higher AI citation rates than comparable sites without it.
Add this JSON-LD block to any content page you want Perplexity to treat as current:
Update dateModified only when you make a real content change: new statistics, a corrected claim, an added section, or a removed outdated reference. Changing dateModified without changing the page body tells AI crawlers the date is unreliable. Both changes together tell the system the content is current and verified.
OpenAI retires o3 from ChatGPT on August 26: The o3 model leaves ChatGPT's model picker today following a 90-day sunset period. o3-pro stays available for Pro, Team, Enterprise, and Edu subscribers. GPT-5.6 Sol is OpenAI's published replacement for reasoning tasks. The API is unaffected. (OpenAI Help Center)
Google AI Mode now uses Gemini 3.7 Flash for some queries: Google's Search AI Mode has started routing some queries to Gemini 3.7 Flash, launched August 13. Citation behavior in AI Mode may shift as the rollout expands. (Search Engine Roundtable)
Google August 2026 spam update completed rollout: Google's third spam update of 2026 finished rolling out Friday. Sites with thin content or no named author bylines may see citation drops across organic results and Google AI Overviews. (Search Engine Roundtable)
Perplexity Computer expands into Microsoft 365: Perplexity's Computer agent now runs inside Word, Excel, PowerPoint, Outlook, and Teams. AI workflows in enterprise productivity tools create new brand-citation surfaces beyond the Perplexity web interface. (SiliconAngle)
Google AI Overviews adds save and simplify features: Google introduced a "save" function letting users save AI Overview results for repeated queries, plus a "simplify language" option. Both rolled out August 15 in the US. (Search Engine Land)
BrandCited's freshness audit checks every key page for a valid dateModified field in Article or WebPage schema, confirms the date corresponds to actual content changes, and flags pages where the modification date is more than 90 days old. The check appears in the Technical score of your AI Visibility Score with a severity rating and a specific fix. Run the check free at brandcited.ai.
1Add `dateModified` to Article schema on every content page today. Copy the JSON-LD block above and set dateModified to the date of your last real content update. This single change makes your freshness signal visible to Perplexity, ChatGPT Browse, and Google AI Overviews.
2Audit every page published before February 2026. Find each blog post, product page, and guide that hasn't been updated in six months. Each one is a citation candidate Perplexity will exclude under last_updated_filter queries.
3Make substantive updates, not cosmetic ones. Add a current statistic, update an outdated claim, or insert a new FAQ entry. AI crawlers compare page snapshots over time. A date change without content change reduces retrieval trust.
4Update your llms.txt file with a last-verified note. If your llms.txt lists key pages, add a date note to each entry. This reinforces freshness signals at the file level that Perplexity reads directly.
5Set a 90-day refresh calendar for high-value pages. Pages not updated within 90 days are three times more likely to drop from AI citation pools. A calendar entry costs nothing.
6Run a BrandCited scan to find which pages fail the freshness check. The audit shows every page with a missing or outdated dateModified field, ranked by how much citation volume that page generates. Fix the highest-traffic pages first.
Perplexity's last_updated_filter is a Search API parameter that restricts results to content modified within a specified time window. When the filter is applied, pages without a valid last-modified timestamp or with timestamps outside the window are excluded from results entirely — not ranked lower, but fully removed. It was added in August 2026 alongside the existing published_after and published_before parameters.
Does dateModified schema affect ChatGPT citations?
ChatGPT uses training data and real-time Browse mode. Article schema with dateModified affects ChatGPT Browse citations directly, where freshness is a ranking signal. For training-data citations, dateModified has less immediate impact, though it helps ensure the current version of content is indexed. Perplexity responds faster to freshness signals because it retrieves content in real time on every query.
How often should content be updated to stay in Perplexity's citation pool?
Update high-value pages at minimum every 90 days. Pages not refreshed within 90 days are three times more likely to drop from AI citation pools, according to Onely's 2026 data. For pages targeting comparison or news-adjacent queries — the types most likely to trigger Perplexity's freshness filter — a 30-day refresh cycle keeps citation rates in the top tier.
Can you just change the dateModified timestamp without updating the content?
No. AI crawlers compare page snapshots over time. A dateModified change without substantive content change reduces retrieval trust rather than improving it, according to Authority Tech's 2026 freshness analysis. Perplexity's system detects cosmetic freshness and discounts it. Update the date only when you've made a real change: new data, an added section, a corrected claim, or a removed outdated reference.
Does content freshness matter equally for all nine AI engines?
No. Perplexity is the most aggressive freshness gater — its citation rate drops from 82% to 37% for content older than 180 days. ChatGPT and Claude are more forgiving of older authoritative content. Google AI Overviews weights freshness for news-adjacent queries but cites older definitive sources for evergreen topics. BrandCited's per-engine audit shows your freshness standing on each of the nine engines separately.
