Google tests AI-generated headlines in Discover — some proved inaccurate

Google is experimenting with AI-generated headlines in the Discover feed: some of the reworded headlines have proved misleading or inaccurate. The service is adding flagged summaries. The company says this is an experiment for a subset of users. Other product changes were also announced.

455
Share:

How headlines are changing

Google is rolling out a test of automatic headline generation using artificial intelligence for the Discover feed. In some publications the new phrasings replace the original headlines; some of these variants have proven misleading or inaccurate.

For example, one article received a headline about an allegedly revealed device price, although the article text did not contain that information. In addition to headlines, the system creates short summaries of pieces; all generated elements carry the label "Generated by artificial intelligence, may contain errors".

Comment and Google products

The company describes the initiative as a small experiment available only to a portion of users.

The update is intended to help better understand topics before clicking through.

– Mellori DeLeon, Google spokesperson

Google also recently clarified free image-creation limits in Nano Banana Pro and granted students in Ukraine a year of free access to AI Pro.

World news