Real-world examples of AI in journalism — built for editors, reporters, and local news teams.
Showing 21 of 1072 in this view
A machine-learning powered dynamic paywall system that uses user behavior and content performance data to determine whether to show paywalls to individual users. The system estimates potential ad revenue per user and adapts paywall tightness based on news cycle activity to optimize both subscription conversions and ad revenue.
Social Card Generator is the tool that two Northwestern students built with Claude to help the Ocala Gazette upgrade its Instagram and Facebook posts. It’s a simple HTML page that runs in a browser: a journalist uploads a photo, types their own headline and text, and the tool instantly renders a branded social card matching the Gazette’s site style (logo, colors, typography). AI is used only to design and refine the layout and interface, not to write or fact‑generate content, so reporters keep full control of wording while gaining a faster, more professional-looking social workflow that takes only a few extra seconds per post.
A newsroom tool designed to detect AI-generated and manipulated content (text, images, video, and audio) before publication. It integrates into content management systems to provide automatic pre-publishing verification and includes compliance reporting features.
An AI tool built by Reach plc that assists journalists in their reporting workflow. The tool was praised by judges for enabling journalists to focus on the stories that matter, winning the best use of AI award at the Future of Media Awards 2026. Originally born out of Reach's "Mantis" contextual intelligence and brand safety ecosystem, Mantis Scribe was introduced to automate internal editorial tasks and streamline article creation workflows.
AgentCrunch runs an AI‑native news magazine in Israel where six autonomous Enso AI agents handle the entire content pipeline. The agents work 24/7 to source, write, edit, fact‑check and publish articles using a custom in‑house language model, with no human editors or writers involved. The workflow delivers a continuously refreshed magazine produced entirely by AI.
Politico is consolidating its previously fragmented publishing systems into a single headless CMS that serves as the core “plumbing” for all content creation and distribution across its US and European operations. By moving to this unified, API-driven platform and pairing it with one global app that surfaces both free and paywalled journalism, Politico can experiment faster with AI—using shared content structures and metadata to power AI-driven search, recommendations, personalization, workflows, and product experiments without constantly rebuilding backend systems for each newsroom or product.
444.hu has deployed the open‑source REMP (Reader’s Engagement and Monetization Platform) together with its built‑in machine‑learning churn‑prediction module. The system analyzes subscriber behavior, forecasts likely cancellations and feeds those insights into the newsroom’s revenue‑analytics workflow, allowing editors and sales teams to prioritize retention actions. By combining AI‑assisted forecasts with human decision‑making, the publisher aims to improve its subscription stability and overall monetization performance.
A Lincolnshire local-news operation built around an AI journalist called Cathy Lindum. The site's own description says Cathy scans the web for news, events and developments throughout Lincolnshire and produces the initial journalism. The Lincolner explicitly says every Cathy-generated piece receives human editing. Editors fact-check and improve the stories before publication.
Al Jazeera Media Network uses its in‑house “Listen‑to‑article” feature to automatically convert selected news stories and opinion pieces into audio with a custom AI text‑to‑speech model. The audio version is generated and published moments after the written article goes live, giving readers a hands‑free option and boosting accessibility for on‑the‑go consumption. The process runs fully automated, requiring no human intervention beyond the original article creation.
A regional news publication covering Boyacá, Colombia, which describes itself as the country's first newspaper operated by a newsroom of AI agents under human editorial supervision. Instead of one AI reporter, Boyacá Digital describes a chain of specialized agents: Editor-in-chief → Reporter → Writer → Fact checker → Copy editor → Legal guardian → Art director. The editor sets the agenda; the reporter gathers and attributes sources; the writer produces the article; the verifier checks claims; the copy editor checks style/neutrality; and the legal agent can stop a story over risks such as defamation or personal-data issues. A human editor remains responsible for the publication.
An AI-native publication covering the AI, startup and technology economy, founded by Ryan Merket. It is unusually close to a one-person newsroom: Merket built the system and supervises it, while AI agents perform much of the reporting, curation, writing and publishing. RuntimeWire describes its primary product as document-led investigations, technical research, exclusives and scoops, with a continuous news feed alongside it.
BBC's provenance tool links journalists’ claims directly to verified source documents, such as government press releases, using AI and cryptography. Canonical sources are first captured as high-fidelity, tamper-evident web archives (WACZ files). An AI Claim Matching Algorithm then extracts and semantically matches key claims from these sources to passages in news stories, even when they’re paraphrased, and a journalist confirms the match. This creates a two-way, cryptographically sealed attribution link, so readers can click from a claim back to the original document and see if it has changed or been deleted, triggering a visible “broken seal” warning if something no longer aligns. The goal is to make it trivially easy for anyone to verify who said what, when, and in what context, strengthening trust in text-based news in an AI-saturated information environment.
Bauer Media’s AI & Tech Initiative is an integrated “liquid publishing” system that treats content as structured, modular knowledge rather than finished articles, and embeds AI across the entire lifecycle — from trend analysis and research to briefings, drafting, optimisation, publishing, and lifecycle management. Instead of separate chat tools, AI is woven directly into editorial workflows as an intelligent layer that observes work, prepares inputs, and automates low‑value tasks while preserving human editorial control and transparency over how AI is used. The result is a workflow-first, AI-first publishing platform that reduces manual effort, improves speed and quality, and forces the organisation to align content structure, product, and commercial logic into one coherent operating model.
Tempo is building a combined Digital Asset Management system and AI chatbot (“Tempo Assistant”) that turns more than 50 years of its archives into a searchable knowledge assistant for journalists and readers. All photos, articles, videos, audio, and documents are centralised in one DAM, enriched with metadata and contextual search (people, places, time, events) so reporters can quickly find and reuse material. On top of this, Tempo Assistant uses a Retrieval-Augmented Generation setup where an LLM answers user questions only from Tempo’s verified archives, always linking back to original articles and refusing to invent facts when sources are missing. The goal is to cut newsroom drudge work, give audiences conversational access to Tempo’s reporting, and ultimately evolve the archive into a knowledge graph and service that can be extended to other organizations.
Handelsblatt's editor-in-chief Sebastian Matthes described an in-house-developed agent system now used across the newsroom for research support, feedback on drafts and preparation for interviews, alongside general-purpose use of Claude for developing product ideas and editorial strategy. The tool reflects a broader stance Matthes has articulated for the German business daily: AI can assist and improve journalism but final authorship and responsibility for published text remains with the human writer.
MediaScoper is a Narrative Intelligence Console for OSINT-style media monitoring aimed at civil society, fact-checkers, threat analysts and journalists. It collects and analyzes streams from news sites and hundreds of Telegram channels to visualize how narratives emerge, spread and fade before they go viral, supporting detection of coordinated disinformation (FIMI), polarization analysis and spike alerts that replace manual channel monitoring. The stack uses Python scraping, Elasticsearch, exact/semantic/hybrid search, local LLMs via Ollama for summarization, entity extraction and sentiment, a React frontend, and Cloudflare protection. The product evolved from an internal news aggregator into a cloud SaaS with private dashboards and PDF/CSV report export.
Hamilton is the New York Post’s new AI-powered engagement platform that sits inside its app and, soon, website to make the Post’s journalism more discoverable, personalized, and habit-forming while tightly controlling AI costs. It combines conversational search over five years of archives, grounded and cited answers, personalized recommendations, time-sensitive AI news digests, and one-to-one push notifications tuned to each reader’s behavior. Under the hood, its most important feature is a caching layer that stores answers to common questions so the Post doesn’t pay for repeated AI calls, with real-time monitoring of API usage and repeat queries to keep spending sustainable as adoption grows.
This tracker monitors developments between news publishers and AI companies—including Lawsuits, Deals, and Grants—based on publicly available information.
GuIA (an abbreviation playing on "IA avanzada para el sistema editorial") is an AI system that El País has integrated into its content management system to automate mechanical editorial tasks. It handles automatic content tagging, selection of related news, and generation of summaries, aiming to cut operational load on the newsroom so journalists can spend more time on reporting and content creation. The project is co-financed by Spain's Secretariat of State for Digitalization and Artificial Intelligence together with EU Next Generation funds, under Spain's AI-in-media grant program and the España Digital 2026 plan.
Nieuwslab is a one-person experimental “AI newsroom” where journalist Melle Drenthe lets AI handle research and draft-writing for every article while he acts as the final human editor, technically required to review, adjust, and approve each piece before publication; the entire site itself was designed and built by AI under his direction, and the project systematically documents both the successes and failures of this workflow to explore where AI can meaningfully support journalistic work and where it clashes with professional standards, ethics, and craft.
Ampop Smart Flow is an AI news-generation workflow for the Ampop newsroom that produces draft materials meeting Ampop’s editorial standards. Its goal is to speed up Ampop’s process by filtering daily information from nine sources and automatically assembling stories in the explanatory journalism genre. The documented tech stack centers on Google Gemini.
Showing 21 of 1072 in this snapshot · 1072 total in the live archive.