No one wrote this story

By Joshua Vickers

Once gatekeepers of public discourse, journalists now compete with code for readers’ attention and trust.

Two voices stood out at TED in Vancouver earlier this month. Carole Cadwalladr, the British journalist who exposed the Facebook-Cambridge Analytica scandal, warned of a “technological coup” that risks dismantling democracy. Sam Altman, the CEO of OpenAI, predicted AI would become humanity’s most transformative invention—but conceded it was “crazy not to be a little bit afraid.” Between these poles of dystopia and techno-optimism lies the newsroom, where journalists are discovering that the AI future is neither entirely robotic nor entirely reassuring.

Legacy media is being transformed. Artificial intelligence is now a regular newsroom contributor—though it never asks for a byline. It writes summaries, tags articles, suggests headlines, and, in some cases, drafts entire pieces. But even as it changes how journalism is produced, AI is clarifying something more fundamental: readers trust humans more. And they trust human judgement most of all.

The Bots Are Filing. The Editors Are Watching.

Newsquest now employs 36 ‘AI-assisted reporters,’ up from just seven in late 2023. These journalists feed press releases into algorithmic engines that churn out clean copy for human review. One Reach reporter recorded 150 bylines in a single day. Bild in Germany has downsized editorial staff in favour of automation. Gannett in the US now advertises for AI-powered sports reporters who never leave their desks. Publishers maintain that these tools free up reporters to chase real stories. But even insiders express unease. One UK subeditor called AI a “blessing and a curse”—it saves time, yes, but erodes authorship. A regional journalist noted that AI handles the boilerplate well, but has “gutted the sense of craft”. In January 2025 alone, over 900 editorial jobs vanished in the UK and US. Efficiency is rarely unaccompanied by redundancy.

A Global Divergence in Strategy

Britain has responded to AI with pragmatic experimentation. The Guardian licenses content to OpenAI, and Reach uses AI summaries to hook readers. Regulators are watching but not yet intervening.

In contrast, the United States is characteristically polarised. News Corp has inked a $250m licensing deal with OpenAI, and the New York Times is suing. In Europe, regulators are reaching for the red pen. France’s data watchdog has launched investigations, and the EU’s AI Act may soon require consent for content training.

Platforms Reap. Publishers Reckon.

Google’s AI-powered search tools now answer queries directly, often using journalistic content scraped and summarised. Some UK publishers have seen traffic fall by as much as 70%. The founder of World History Encyclopedia described the new AI powered internet as a deal no longer mutual. Publishers provide the raw material but get neither credit nor clicks.

This disintermediation has existential implications. Fewer visits mean lower ad revenue and weaker subscriber pipelines. Publishers are doubling down on loyalty, nudging readers into apps, games, and newsletters. At The Times, churn plummets when readers visit ten times a month. Frequency is becoming a proxy for financial survival.

Licensing or Litigating: Pick a Side!

Some publishers are making deals. The Guardian’s OpenAI partnership brings reliable content into ChatGPT, with attribution and compensation. Insiders estimate such deals could be worth tens of millions. Others, like The New York Times, have opted for the courtroom.

This divide reflects a deeper uncertainty: is journalism a public service or a licensable product? The answer, increasingly, is both—depending on whether AI models ask for permission or forgiveness.

Routine Rewritten. Judgement Remains.

AI excels at translation, transcription, and templated stories. Bloomberg and the FT use it to scan datasets. News Corp Australia publishes thousands of AI-generated local briefs. But trust remains elusive. A YouGov poll found only 13% of Britons trust AI-written articles. If both author and editor are algorithms, 72% express active distrust. Transparency helps—BuzzFeed and Reach label AI content—but also hurts: when readers know, they doubt. AI is efficient, but it is not accountable. It cannot explain why a story matters, or what angle is missing. It can describe what happened. It cannot decide what deserves to be told.

Cadwalladr and Altman: Redux

If Cadwalladr is right, AI could become an unaccountable force that drowns journalism in noise. If Altman is right, it could amplify good journalism and help it reach farther, faster. But both agree on one point—the risk is real.

Paradoxically, that risk is now journalism’s opportunity. The flood of synthetic content has created scarcity where it matters most: trust. Human journalism, with all its flaws and expense, has never been more valuable—not for its speed—but for its judgement. Legacy media will survive this disruption as it has others. But only if it remembers what its readers already know: information is easy. Understanding is rare. And the byline—still human, still accountable—is journalism’s last line of defence.

Image: Joshua Vickers