What it is
In June 2026 the IETF (the body behind HTTP, DNS, and most of the internet’s low-level standards) published RFC 10008, adding a new standard HTTP method: QUERY. It resolves an old dilemma. When a program runs a complex query against a server (think a product filter with dozens of conditions, or a cross-field report query), GET forces every condition into the URL — long ones break, and sensitive filters linger in browser history and server logs. Switching to POST moves the conditions into the request body, but POST semantically means “write”: clients won’t auto-retry after a dropped connection, and caches skip the response by default. QUERY takes the best of both: read-only, retryable, and cacheable like GET, while carrying a full query body like POST.
The supporting pieces ship with it: servers can advertise the query formats they accept via the Accept-Query header, unrecognized formats get a 415, well-formed but invalid queries get a 422, and cross-origin use always passes through the browser’s preflight check, so it opens no new security hole.
What it means for you
The conclusion first: your site gains no AI visibility from QUERY today, and loses none. When ChatGPT, Perplexity, and other mainstream AI engines fetch your pages and decide whether to cite you, they still use plain GET. QUERY governs how programs run structured queries against APIs — it says nothing about how AI reads your content.
The reason to remember it comes next. AI is moving from reading pages toward querying data directly: agents checking stock, comparing prices, booking tables on a user’s behalf. QUERY gives that kind of query a safe, cacheable, standard semantics — plumbing for that stage of the web. When AI engines start requesting data from sites through structured queries, sites with standard interfaces benefit first. That is an outlook, not the present.
Do you need to act
No. Content sites, brand sites, and storefronts have nothing to change. Caches must fold the query body into their cache keys before any caching benefit materializes, and framework and infrastructure support is only getting started. Engineering teams running their own APIs can start paying attention; everyone else should keep their effort where AI actually reads today — content and structure. Standards will keep arriving; sorting out which ones affect your AI visibility and which stay engineering-only is part of what we watch for our clients.