<time datetime=""> used for dates
What it checks
AI agents use <time datetime> elements to reliably parse dates for freshness scoring and temporal reasoning. Without machine-readable dates, agents must regex-parse human-readable date formats, which frequently fails across locales and ambiguous formats like “01/02/2025”.
Why it matters
A <time datetime="YYYY-MM-DD"> exposes the publication or modification date in an unambiguous ISO-8601 form, so date-extraction libraries embedded in LLM corpus-building and RAG pipelines resolve the date deterministically instead of pattern-matching ambiguous prose (‘03/04/25’, ‘last Tuesday’). Pages without it force those pipelines onto text heuristics that are locale-ambiguous and frequently wrong.
Evidence
time element with datetime attribute as a machine-readable…
Ratified standard with a clear intent: ‘The time element represents its contents, along with a machine-readable form of those contents in the datetime attribute’ [whatwg-time-element]. There is at least one real consumer chain. htmldate’s fast-path selector explicitly includes the <time> element as a date-extraction target [htmldate-extractors]. htmldate is peer-reviewed, and reported as running ‘in production on millions of documents’ [htmldate-joss-paper]. And trafilatura, the standard main-text extractor for LLM web corpora, uses it for metadata [trafilatura-corefunctions].
Limits
Substantial, and it caps this at C. (1) HTML-AAM maps <time> to the GENERIC role [w3c-html-aam], meaning the element and its datetime attribute are invisible in the accessibility tree — so Playwright MCP, Chrome DevTools MCP, Anthropic read_page and every other a11y-tree agent never see it. (2) Google’s own publication-dates guidance recommends exactly two things, a prominent user-visible date and structured-data datePublished/dateModified, and does not mention <time datetime> [google-publication-dates]. (3) The WHATWG spec defines parsing syntaxes but prescribes no user-agent behaviour for consuming the value [whatwg-time-element]. (4) Even htmldate keys on the element as a text container — its constants match @itemprop and date-ish @class/@id, with no XPath on @datetime itself [htmldate-extractors]. (5) HTML→markdown conversion drops the attribute.
The A-grade path to a machine-readable date is JSON-LD datePublished, which belongs to the structured-data domain; <time> is a cheap complement, not a scored requirement.
Sources
- HTML Standard — the time element — WHATWG, spec (verified 2026-08-20)
- htmldate extractors.py (date XPath constants) — Adrien Barbaresi / htmldate, vendor-doc (verified 2026-08-20)
- htmldate: A Python package to extract publication dates from web pages — Journal of Open Source Software (Barbaresi, 2020, 5(51):2439), study (verified 2026-08-20)
- trafilatura core functions documentation — Adrien Barbaresi / trafilatura, vendor-doc (verified 2026-08-21)
- HTML Accessibility API Mappings 1.0 — W3C, draft-spec (verified 2026-08-21)
- Add a Byline Date to Google Search Results — Google Search Central, vendor-doc (verified 2026-08-21)