Explainer
Will AI Replace Journalists?
The honest version of the answer, from a newsroom that runs on automation: what AI already does in news, what it cannot do, and what changes next.
By the AI2Day Newsdesk · Updated 23 July 2026 · A living guide, revised as the industry moves
In short
- AI is replacing journalistic tasks, not the journalistic role: formats with fixed shapes (market wraps, summaries, translations) automate first, while reporting that requires sources, judgement and accountability stays human.
- AI news anchors and AI-written articles already exist at scale, which changes newsroom economics and job descriptions rather than eliminating the profession outright.
- The realistic future is smaller newsrooms doing more with automation underneath, and the interesting question is who owns the standards. We would say that: this site is an automated newsroom with humans owning the standards.
Will AI replace journalists?
AI is replacing tasks inside journalism faster than it is replacing journalists. Formats with a fixed shape (market reports, sports recaps, translations, summaries of official releases) automate readily and already have. Original reporting, source cultivation, verification under pressure and accountability for being wrong remain stubbornly human, and every serious newsroom experiment has rediscovered that line.
The honest framing is economic rather than existential. Automation lowers the cost of the repeatable layer of news, which shifts human effort toward the parts that cannot be templated: finding what nobody has published, getting people to talk, judging what matters, and standing behind mistakes. Industry surveys, including the Reuters Institute's annual research, consistently find newsroom leaders planning AI into production while keeping editorial judgement with people. Jobs change shape before they disappear, and the shape is already changing.
How do newsrooms actually use AI today?
Far more than most readers realise, and mostly below the byline: transcription, translation, first drafts of templated stories, headline testing, archive search, and monitoring feeds for something worth a human's attention. What readers picture (a machine inventing the whole paper) is the rare case; the assembly line underneath is the common one.
The phrase newsroom AI usually means that assembly line. Wire agencies have auto-generated earnings and sports stories for years; international outlets machine-translate their own reporting to reach new language markets; investigative teams use models to search documents at a scale no intern could. There are also newsrooms, this one included, where automation runs the pipeline end to end within human-owned standards: AI2Day stories are researched from primary and established sources, rewritten for a mainstream reader, checked automatically, and published around the clock with the whole process documented. Agentic systems (software that monitors, drafts and files with a person supervising) are the current frontier, and our frontier labs section tracks the models driving them.
What is an AI news anchor?
An AI news anchor is a generated on-screen presenter reading real (or supposedly real) news: a synthetic face, a synthetic voice, and a script from somewhere. The first mainstream example launched in China in 2018, India followed with several Hindi and regional-language anchors from 2023, and the format has since spread widely.
Anchors automate presentation, which was always the most templated part of broadcast news, so the substitution is less radical than it looks. The journalism, if any, still happens upstream in the script. That is also the risk: a synthetic presenter reads propaganda exactly as fluently as news, and several disinformation campaigns have used fake anchors for false stories precisely because the format borrows television's authority. The sensible reader response is the same one from our reliability guide: judge the outlet's sourcing and accountability, not the face reading the autocue.
Can AI write news articles?
Yes, mechanically well and journalistically incompletely. Given sourced facts, modern models draft clean, accurate copy in seconds. What they cannot do is the part before the draft: witness events, work sources, extract documents nobody handed over, or take responsibility when a claim is wrong.
That split explains both the successes and the failures you have read about. Automation on top of solid inputs (wire feeds, filings, verified announcements) produces reliable stories at scale, which is exactly the lane this site operates in. Automation asked to know things it was never given produces confident fabrication, and outlets that published unchecked model output have collected corrections and ridicule for it. The craft has moved: the scarce skill is designing and supervising pipelines where every claim traces to a source, plus the reporting itself, which remains the one input machines cannot generate. Our guide to AI and the future of humans covers the same task-versus-role pattern across other professions.
Could AI save the news business?
It could save the economics of covering news; it cannot by itself restore trust or replace the reporting layer. Automation cuts production costs dramatically, which makes niche and around-the-clock coverage viable again, including in places that lost their outlets entirely. Whether that revives local news depends on who runs the pipelines and to what standard.
The collapse of local journalism was economic: the cost of a newsroom outran the revenue of a small audience. Automation attacks exactly that cost, and a two-person outlet with a good pipeline can now sustain coverage that needed a dozen staff before. The essays asking whether AI will save or finish the news tend to agree on the mechanics and split on the ethics, and both outcomes are on the table: the same tooling powers careful niche outlets and zero-standards content mills. Our bet, stated plainly because we are one of the test cases, is that automation plus published standards plus human accountability is a sustainable model for beat coverage. The future of AI guide holds the wider forecast.
Frequently asked questions
Will AI replace news anchors and presenters?
Presentation is the most automatable layer of broadcast news, so synthetic anchors will keep spreading, especially for bulletins and translations. Presenters whose value is interviewing, judgement and trust are far harder to substitute than presenters who read an autocue.
Will AI replace news reporters and writers?
Writers of templated copy are already being displaced; reporters who generate original information are not. The distinction is the work before the writing: sources, verification and accountability. Roles built on those keep their value as the writing layer automates.
How many journalism jobs will actually be lost?
Nobody has a trustworthy number, and headline forecasts have aged badly in both directions. What is observable now: newsroom production roles are shrinking, pipeline and editing roles are changing shape, and outlets are publishing more output per journalist employed.
Should journalists learn AI tools?
Yes, in the same way they once learned the phone and the search engine. Journalists who supervise pipelines, verify at scale and design automated coverage multiply their reach; those who compete against the tools on speed of templated copy are competing where the machine is strongest.
Does AI2Day employ journalists?
AI2Day is an automated newsroom: software produces the stories from primary and established sources, and people own the standards, corrections and direction. That makes us part of this story, which is why this page states its position openly rather than pretending to neutrality.
Watch it happen
The AI and journalism story is moving weekly: anchors, newsroom deals, model releases and the occasional public failure. We cover it as it lands.