What Google and Bing actually say about schema and AI

Two search companies, two different public positions on the same question. Google’s developer documentation says plainly that structured data is not required to appear in AI Overviews or AI Mode. Microsoft has been reported, separately, as saying schema markup helps its Copilot systems understand content. Neither statement is the sweeping “schema equals AI visibility” story that circulates in SEO blogs, and schema.org’s own release notes show the vocabulary keeps growing for reasons that have little to do with AI search specifically. Here is what each party has actually said, and what has not been verified.

What schema.org’s own numbers say#

Schema.org’s documentation states the vocabulary “currently consists of 823 Types, 1,529 Properties, 19 Datatypes, 96 Enumerations and 535 Enumeration members.” The most recent major release, version 30.0, shipped on 19 March 2026 and added a Credential class and an Error class with an errorCode property, among other additions logged in the project’s own release notes. That is a durable fact anyone can check today by opening the release history. It is not, on its own, evidence that AI systems weight structured data more heavily than they did before the release.

What Google says, and does not say#

Google’s “AI features and your website” documentation, last updated 10 December 2025, addresses the structured-data question directly: “There’s also no special schema.org structured data that you need to add” to appear in AI Overviews or AI Mode. The same page still recommends existing structured-data practice as part of general SEO, but it stops short of claiming schema markup earns a site AI citations specifically. That distinction, required for rich results eligibility versus optional for AI features, is the one worth keeping straight before adding markup on the strength of an AI promise.

What Bing has said, and what remains unverified#

Search Engine Land and Search Engine Roundtable both reported in March 2025 that Fabrice Canel, principal product manager at Microsoft Bing, confirmed on stage at SMX Munich that schema markup helps Microsoft’s LLMs, including Copilot, understand web content. Neither outlet published a verbatim quote of Canel saying that specific sentence: the confirmation traces back to a LinkedIn post summarising his talk, which Canel then engaged with in the comments rather than issuing his own written statement. Treat the underlying claim, that schema aids comprehension, as credibly sourced but not independently quotable, and treat any number attached to it (a lift percentage, a citation-rate multiplier) as unverified until a named study says otherwise.

Why we care#

The honest version of this story is smaller than the one making the rounds: one company says structured data is optional for AI features and always has been, the other has a reported, not directly quoted, statement that it helps their models understand a page. Neither has published a number connecting schema completeness to citation frequency. A publication that repeats an unsourced lift statistic because it sounds plausible has stopped doing the one thing that separates reporting from marketing copy, tracing every figure back to where it came from.