Google Tests AI Headlines in Search After Discover Trial
▼ Summary
– Google is testing AI-generated headline rewrites in its Search results, describing the test as “small and narrow.”
– The AI rewrites can create new phrasing not found in the original article, differing from older rule-based title changes.
– Google’s stated goal is to match titles better to user queries to increase engagement with web content.
– Publishers express concern over losing control of their brand voice and potentially misleading readers without disclosure.
– There is currently no publisher opt-out or tool to detect these rewrites, requiring manual checks of search results.
Google is now experimenting with AI to rewrite headlines in its main search results, a move that mirrors its recent approach in the Discover feed. The company confirmed a limited test aimed at creating titles that more closely match user queries and encourage engagement. This development follows a similar pattern from late last year, when Google initially described AI headlines in Discover as a small test before quickly reclassifying it as an official feature.
Several examples of these AI-generated headline rewrites have been observed in search listings over recent months. In one instance, an article titled “I used the ‘cheat on everything’ AI tool and it didn’t help me cheat on anything” appeared in results as simply “‘Cheat on everything’ AI tool.” Another was altered to “Copilot Changes: Marketing Teams at it Again,” a phrase not found in the original content. Google states the test affects various website types, not just news publishers, and currently does not include any disclosure that a headline has been changed.
The company’s stated goal is to identify the most useful and relevant page title for a given search. While this test employs generative AI technology, Google notes any future, broader implementation might not, though it provided no details on alternatives. For now, the experiment remains unapproved for a wider rollout.
This new test represents a shift from Google’s long-standing practice of title tag rewrites. For years, the search engine has used rule-based systems to modify page titles in results, with studies showing it changes a majority of them. However, those traditional rewrites pull text directly from existing on-page elements like H1 tags or meta descriptions. The current AI experiment is distinct because it can generate completely new phrasing, as seen in the Copilot example.
The potential expansion of this feature from Discover to core Search carries significant weight for publishers. An analysis indicates Discover’s share of Google traffic to publishers has grown substantially, meaning many sites now rely heavily on both Discover and Search. Losing headline control across these two primary channels could impact audience reach and brand voice. Google’s documentation outlines what page elements it may use for titles but does not offer publishers a way to opt out of rewrites. Since changes are not disclosed, publishers must manually check their listings to see if their headlines are being altered.
Industry reactions highlight deep concerns. A senior editor likened the practice to a bookstore replacing book covers and titles, while an SEO director emphasized that headlines are crucial for attracting readers and building long-term audience trust, which can be compromised if facts are misrepresented.
For content creators, vigilance is currently the only option. Monitoring how article titles appear in Google search results requires manual spot-checking, as no dedicated tool exists for this purpose. The progression from test to feature in Discover suggests this search experiment could follow a similar, rapid path, making it a critical issue for the publishing ecosystem to watch.
(Source: Search Engine Journal)
