Google AI changes news headlines! You may be misled without knowing it

Google is testing an AI feature in its Discover news feed that automatically rewrites headlines, aiming to help users grasp articles more quickly. However, early examples show the AI-generated headlines can be misleading, oversimplified, or factually incorrect.

For example, an Ars Technica story originally titled “Valve’s Steam Machine looks like a console, but don’t expect it to be priced like one” was rewritten by AI as “Steam Machine price revealed”, which is inaccurate. Similarly, a report on Radeon GPUs titled “Radeon RX 9070 XT Outsells The Entire NVIDIA RTX 50 Series On Popular German Retailer” was shortened to “AMD GPU tops Nvidia”, stripping away critical context. Other rewrites, such as “Schedule 1 farming backup”, have been unintelligible, replacing otherwise clear and detailed headlines.

While Google says the experiment affects only a small subset of Discover users and is intended to make topic details easier to digest, the AI headlines carry a disclaimer—“Generated with AI, which can make mistakes”—that is only visible if readers click “See more.”

Critics warn that replacing carefully crafted editorial headlines with AI-generated versions undermines editorial control, misleads readers, and risks reducing trust in journalism. By prioritising brevity or engagement over accuracy, the AI rewrites could effectively create clickbait, even though Google’s own guidelines prohibit it.

The experiment also highlights broader concerns about how tech platforms shape news consumption. Experts emphasise the need for transparency and accountability, warning that AI should not compromise the integrity of reporting or the careful judgement of human editors.

As AI continues to influence digital media, the challenge remains to balance convenience and clarity for users with the accuracy, context, and credibility that form the foundation of trustworthy journalism.

Google is reportedly conducting a small-scale experiment in its Discover news feed that uses AI to automatically rewrite news headlines. The company states the goal is to help users quickly understand article topics. However, the AI-generated headlines have sometimes been inaccurate, misleading, or unintelligible.

Examples of problematic rewrites:

• Ars Technica’s original: “Valve’s Steam Machine looks like a console, but don’t expect it to be priced like one” → AI: “Steam Machine price revealed” (factually incorrect)
• Radeon GPU report: “Radeon RX 9070 XT Outsells The Entire NVIDIA RTX 50 Series On Popular German Retailer” → AI: “AMD GPU tops Nvidia” (loss of context)
• PC Gamer story: “Schedule 1 creator had a backup plan if Steam rejected it…” → AI: “Schedule 1 farming backup” (unintelligible)

The AI-generated headlines carry a small disclaimer (“Generated with AI, which can make mistakes”), but this is only visible if users tap “See more,” meaning most readers may not realise the changes were made by AI.

Concerns raised:

• Editorial control is undermined as carefully crafted human-written headlines are replaced.
• Misleading or simplified headlines can confuse readers or be perceived as clickbait.
• Publications could suffer reduced readership or trust, affecting revenue.
• Transparency and accountability are critical, as AI intervention could distort how news is presented.

The experiment underscores a larger debate about the role of AI in news distribution, balancing user convenience with maintaining journalistic integrity. Experts emphasise that AI should not compromise the accuracy, context, or reliability of reporting, which remain essential for trust in media.