Google Trials AI Headline Rewrites in Search After Discover Rollout

Google is once again experimenting with the way how headlines appear on its platforms. After testing AI-generated headlines in Discover initially and quickly turning them into a feature, the company is now trying something similar in Search.

The move has caught the attention of publishers, as headlines are central to both traffic and brand identity.

A Familiar Test Pattern Emerges

The company has confirmed that it is running a limited test of AI-generated headline rewrites in Search. The phrasing used to describe the test is notable. Google calls it “small and narrow,” the same wording it used during the early days of its Discover experiment.

That earlier test moved quickly. Within about a month, what started as an experiment became a full feature in Discover. The similarity in language suggests a pattern that publishers are watching closely.

What Users Are Seeing in Search?

Several instances of rewritten headlines have been seen to have already surfaced. In some cases, original titles have been shortened or altered entirely to match what Google believes is relevant to a user’s query when looking for a specific term. In others, the rewritten headline introduces a specific wording that does not appear anywhere in the article itself.

This signals a shift. Traditional title rewrites relied on existing page elements. The new test appears to go further by generating entirely new phrasing using AI.

The changes are not limited to news content. Google has indicated that the test applies across different types of websites. Another key detail is the lack of transparency. Users are not informed when a headline has been rewritten.

According to Google, the goal is to make titles more useful and relevant. The company says it wants to better match search queries and improve engagement with content.

From Rules to AI Generation

Headline rewriting is not new in Search. Google has been adjusting title tags for years using rule-based systems. Studies have shown that a large share of titles is modified before being displayed in results.

These earlier systems relied on structured signals. They pulled from title tags, headings, and other on-page elements. The output stayed close to the original content.

The current test marks a departure. AI-generated rewrites can introduce new language that is not present on the page. This raises questions about accuracy and editorial control.

Why Publishers Are Concerned?

For publishers, headlines are more than just labels. They shape how stories are understood and how audiences respond. Any change to that wording can affect both trust and click-through rates.

Traffic patterns add to the concern. Discover has become a major source of visibility for many publishers. Its own share of Google-driven traffic has grown sharply and massively in the recent years.

If AI headline rewrites expand in Search as well, the publishers may lose control across two key channels at once and that would be difficult.

There is also no clear way to opt out. Google’s documentation explains how titles are generated but does not offer a setting to prevent rewrites. Since changes are not disclosed, many publishers may not even realize their headlines have been altered.

What Comes Next?

For now, the test remains limited. Google has not approved a wider rollout and has suggested that any future version may not rely entirely on generative AI. Details on alternatives remain unclear.

Publishers are left to monitor the situation manually. Checking how headlines appear in search results is currently the only way to track changes.

The bigger question is whether this test will follow the same path as Discover. If it does, AI-generated headlines could soon become a standard part of Search.