About AI News

Independent news on artificial intelligence

Artificial intelligence is changing how people work, learn, create and govern, and it arrives wrapped in more promises and more fear than almost any technology before it. AI News exists to report what is actually happening: what new systems can and can’t do, who is building them and who is paying, what the research shows, how governments and courts are responding, and what all of it means for the people who have to live with the results.

We aim to be accurate first, clear second and fast third. Every article is grounded in published reporting and primary sources (or, for our video reviews, in the creators’ own videos), links to the sources it draws on, and follows our published editorial standards.

AI News is independent and non-partisan. We are not owned by, funded by or affiliated with any AI company, investor, industry group, political party or campaign. We don’t cheerlead for AI and we don’t campaign against it. We test claims against the evidence.

• Claims are attributed. When a company, a researcher, a government, a booster or a doomsayer says what a system can do, what it will do or how dangerous it is, we report it as their claim, and say whether anyone independent has checked it.
• Announcements are not achievements. We distinguish announced from released from generally available, a preprint from a peer-reviewed paper, and a demo from a product.
• Plain language. We don’t describe software as if it thinks, wants or feels, and we leave words like “revolutionary” and “terrifying” to the people we quote.
• Every side, in its own words. On contested questions of policy and safety we report what each side actually said, accurately and in context.
• One exception: creators’ reviews. Our articles on YouTube creators’ reviews of AI models relay a named creator’s view of a model in the creator’s own spirit, without an assessment of ours, and are labeled as such. The rules above don’t apply to them (see Video reviews).

Our news desks each cover one beat, and every article carries a label telling you whether it is news, analysis, an explainer or opinion.

• ModelsNew AI models and products, what they can really do, and how they compare.
• BusinessFunding, deals, earnings, start-ups and the money behind the AI boom.
• PolicyRegulation, lawsuits, copyright and how governments are handling AI.
• ResearchPapers, breakthroughs and AI in science and medicine, explained without the hype.
• Chips & ComputeChips, data centres, energy and the race to build AI's infrastructure.
• Safety & EthicsRisks, alignment, misuse, deepfakes, bias and the people trying to make AI safe.
• Work & SocietyJobs, education, culture and how AI is changing everyday life.

Opinion

Our opinion section argues clear positions on the questions AI raises, openly and grounded in evidence, never along party lines. Opinion pieces run under named columns rather than under the names of individual writers, and they are always labeled as opinion:

• The Argument — The case, made plainly, on the AI debate of the day.
• Reality Check — Testing the claims of AI companies, boosters and doomsayers against the evidence.
• The Human Factor — What AI means for the people who work, learn and live with it.
• The Editorial Board — The considered view of the publication on the AI questions that matter most.

AI News is an automated newsroom. Our articles are researched, written and checked by AI models, working to rules set by the publisher and enforced in software. No reporter writes them. We think a publication that covers AI should be plain about using it.

1. 1.Find

   The system follows the published feeds of news organizations, specialist trade publications, research institutions and AI companies, and groups reports about the same story. Only stories about AI are considered. A story reported by a single source needs an established newsroom or trade publication behind it, so a company’s own announcement is never enough on its own. Separately, it follows the reviews of a list of YouTube creators who test AI models (see Video reviews).

2. 2.Research and write

   An AI model researches the story in the source reporting and other published material, then writes a news story, analysis, explainer or opinion piece. Every fact must come from that material, never from the model’s own memory.

3. 3.Check

   An automated fact-check tests every factual claim against the sources. Quotations are matched to their source word for word, text copied from a source is caught, and the headline must be supported by the article. Drafts that fail are revised or discarded.

4. 4.Publish and correct

   Articles that pass these checks are published with a linked list of their sources. The publisher oversees the system and can hold, correct or withdraw any article. Corrections are noted on the article and on our corrections page.

Automated checks catch a great deal, but not everything. If you spot a mistake, please tell us. Our editorial standards set out every rule in full.

Some of our coverage of AI models reports reviews by YouTube creators: channels that put new models through their own tests and publish what they think of them. We follow a list of channels chosen by the publisher, and editors can also add a review from one of those channels that our checks missed, such as an older one. When a creator reviews a model, we may publish an article on their review. When several have tested a model the same way (the same kind of task, against the same rival model when they compared it with one), we may publish a roundup that sets their concrete results side by side; reviews that each test something different don’t make a roundup.

These articles work differently from the rest of our journalism. They relay the creator’s view in the creator’s own spirit: when a review is enthusiastic, so is the article, and when the creator is disappointed, so is the article. We don’t balance their verdict, add our own critique or caveats, or set it against benchmarks, other coverage or the model maker’s claims, so the independence rules above don’t apply to them. Each one names the creator in its headline and opening, and carries a note saying whose review it reports and that the opinions are theirs. They are not our assessment of the model.

To do this, our software reads each video’s captions (the text YouTube shows as subtitles, often generated automatically) directly from YouTube. YouTube’s robots.txt disallows automated access to the addresses that serve captions, and its terms of service don’t allow automated access to YouTube without its permission, which we don’t have. We read them anyway, to report accurately on reviews their creators have published for anyone to watch. It is the one place where our crawler doesn’t follow a site’s robots.txt. In return:

• Creators are credited. Every article names the channel whose review it reports and links the videos it draws on.
• Their verdicts stay theirs. The article carries the creator’s verdict, emphasis and tone, criticisms included, as much as they gave them. Their opinions are reported as theirs, never as our own findings.
• Only what the videos say. Everything in the article comes from the videos, apart from the model’s name, maker and release when the creator didn’t give them.
• Quotes are exact. Quotations come word for word from the captions. Automatic captions can mishear names and numbers, so anything that may have been transcribed wrongly is paraphrased instead.
• Sponsorships are disclosed. When a video or its description says it is sponsored, a paid partnership or carries affiliate links, the article says so in a plain sentence.
• Creators can ask us to stop. If you make videos and would rather we didn’t report on them, email [email protected] and we will stop using your videos.

Our editorial standards set out the rules for these articles in full.

The AI models that produce our journalism are made by Anthropic (its Claude models) and OpenAI (its GPT and image models). Both are companies we cover. We pay for their models like any other customer. Neither company owns, funds or sponsors AI News, and neither has any say over, or advance sight of, what we publish.

We cover them exactly as we cover everyone else: no softer, no harsher, and never as promotion. Because a model could be inclined to favor its maker, our facts are checked against outside sources, never against a model’s own account. Articles substantially about Anthropic or OpenAI, or their products, carry a short disclosure note.

AI News is published by AI News Media. It is free to read and is funded by advertising, which is served mainly through Google’s advertising network and may also include clearly labeled sponsorships in our newsletter. Every ad is labeled as an ad.

Advertisers, including AI companies, have no say over what we cover or how we cover it. Ads are placed by automated systems that are entirely separate from the newsroom’s decisions about what to write, and we do not sell sponsored articles or accept payment for coverage. See our advertising standards for the ads we will and won’t accept.

Our crawler identifies itself as InferenceImperiumBot (AINews24Bot before the site was renamed). It reads RSS feeds and article pages to research stories and, for a story about a particular website, may capture its homepage. It also checks the YouTube channels we follow for new videos and reads their captions. It follows robots.txt, with one exception, YouTube (see Video reviews): disallow the inferenceimperiumbot user agent (a rule for ainews24bot still applies) and it won’t fetch your pages. If a story is about your website and we can’t capture it ourselves, because your robots.txt disallows our crawler or opts out of AI use or your site turns automated visitors away, we may instead show the Internet Archive’s most recent capture of your homepage from the past week, credited and linked as such. Like every image we use, it is checked by our AI model before it appears. When we search free photo libraries or download a photo or a video thumbnail, the request identifies itself as InferenceImperium/1.0, with our address and contact email.

We also respect opt-outs from AI use. If your robots.txt disallows any of Anthropic’s crawlers (such as ClaudeBot or Claude-User), or sets Content-Signal: ai-input=no, our crawler will not pass the text of your articles to our AI models. Our models also search the web while researching a story. Publishers we know have opted out are excluded from those searches and are never cited, but that exclusion relies on a list we keep, so if we have missed you, tell us and we will add you. Questions, or a request we haven’t met: [email protected].

Honest about how we work

We tell you plainly that our journalism is produced with AI, and which companies make the models we use.

No invented people

Our bylines are desks, such as the Policy Desk, and named opinion columns, such as The Argument. We never create fictional reporters or columnists.

Quotes are real

Direct quotations appear only when the exact words are found in a source, attributed to the speaker and the outlet that reported them.

Sources you can check

Every article ends with links to the reporting it draws on, so you can read the original for yourself.

Mistakes are corrected in public

Corrections are noted on the article and listed on our corrections page.

Your privacy is respected

Our own analytics use no cookies and store no IP addresses. See our privacy policy.

We read every message. Email is the best way to reach us; our contact page explains what to include so we can help quickly.