Google pitches Genesis AI tool to write news stories

For decades, tech platforms have treated newsrooms as content suppliers. Now, Google wants to sit directly at the reporter's desk and draft the story itself.

Google is actively testing an artificial intelligence tool capable of writing news articles, internally codenamed Genesis. The search giant has already started pitching the product to major news organizations, including The New York Times, The Washington Post, and News Corp, the parent company of The Wall Street Journal.

Pitching an assistant that executives found unsettling

Genesis works by taking in factual information, such as details of breaking events, and turning those notes into finished news copy.

Google pitched the software as a personal helpmate for journalists. The company claims the software is meant to handle repetitive tasks, generate headline options, and offer different writing styles, freeing up reporters to do deeper investigative reporting. A spokesperson compared it to assistive writing features already rolling out inside Gmail and Google Docs, insisting that it is not intended to replace journalists in reporting or fact-checking.

News executives who sat through the pitch were reportedly unnerved. According to The New York Times, two executives described the demonstration as unsettling, noting that the tool seemed to take for granted the intense labor required to produce accurate, artful news stories.

While Google reportedly viewed the project as a way to guide publishers away from the pitfalls of generative AI, the pitch clearly touched a raw nerve inside traditional newsrooms.

The ghost of recent newsroom blunders

Automation in newsrooms is not entirely new. The Associated Press has relied on algorithmic tools for years to churn out quarterly earnings reports. Media outlets like NPR and Insider have also recently told staff they are looking into responsible ways to deploy AI.

However, letting large language models draft actual articles has consistently triggered reputational trainwrecks.

Earlier this year, tech outlet CNET quietly deployed generative AI to produce dozens of personal finance explainers. The experiment backfired spectacularly when readers discovered substantial factual errors and apparent plagiarism across more than half of the published AI pieces. CNET ended up appending lengthy correction notices and assigning staff writers to rewrite the material.

Meanwhile, researchers at NewsGuard recently discovered dozens of automated content farms powered entirely by low-quality AI bots. Newsrooms fear that commercial tools like Genesis could accelerate that degradation of the broader web.

Why it matters

We think Google is making an awkward defensive move here. With OpenAI striking direct licensing deals with The Associated Press and Apple testing its own Ajax-based chatbot internally, Google is desperate to show that its frontier models can serve traditional media rather than destroy it.

However, positioning Genesis as a drafting engine for journalists completely misreads what newsrooms actually need. The bottleneck in serious reporting is almost never the physical act of typing paragraphs; it is sourcing, original verification, and institutional accountability.

Introducing an AI middleman into the copy desk is an enormous operational risk. CNET proved that hunting down subtle hallucinations in AI-generated drafts often consumes more editorial time than writing from scratch.

For publishers, this looks like a trap. Adopting automated writing tools might shave a few minutes off routine wire rewrite duty, but the downside risk is devastating. The moment a synthetic hallucination slips past an editor into print, decades of reader trust vanish overnight.