Skip to content
AI For NewsroomAI Implementations
❤️ Support us

Global knowledge base

AI Implementations

Real-world AI projects, products and experiments from newsrooms around the world — documented in one searchable index.

Index snapshot

1,359Initiatives tracked
932Newsrooms represented
113Countries

Search the index

Find an implementation

Showing 21 of 1359 matching implementations

Or use our free MCP server to work with the database directly in Claude, ChatGPT, Cursor etc. →
01Content Production
Sevenly Magazine

Sevenly Magazine

The magazine of music discoveries, created by the founder of AI For Newsroom. Under the hood, an AI-powered engine uses open data from platforms like Discogs, MusicBrainz, Genius, Last.fm and WhoSampled to uncover connections between songs through songwriters, producers, user likes, genres and more. Other formats help readers discover new music, including City Scene, Producer and Songwriter Spotlights, One-Hit Wonder stories, and a daily music news roundup. All content is produced by AI with human oversight.

AddedOCT 28, 2026
Country index13 tracked
Open implementation ↗Visit newsroom ↗
02Journalism
Projekt Y

Projekt Y

Projekt Y is a new Danish, centre-right news outlet that uses AI as a production tool to vastly scale journalism while keeping human editors firmly in charge. AI systems continuously scan research papers, press releases and other media to surface potential stories and draft articles, but editors still decide what to cover, how to frame it, and what ultimately gets published, adding images at the end. The venture’s goal is to produce more journalism per employee and test real audience appetite for AI-generated news, starting with politics, business and international coverage, and over time evolving into a technology platform others can use to launch their own niche media outlets across areas like sport and culture.

AddedOCT 27, 2026
Country index22 tracked
Open implementation ↗Read coverage ↗Visit newsroom ↗
03Distribution
The Outlaw Ocean Project

Food For Feed

A customized investigative chatbot built by The Outlaw Ocean Project that sits on its website and lets users explore a massive, AI-processed database about the global fishmeal supply chain. On the backend, the team runs open‑weight models like Qwen, Gemma, and Phi on its own hardware to classify unstructured documents, extract entities such as companies and locations, and maintain running summaries on roughly 600,000 vessels, 1,400 factories, and hundreds of importers and retailers. On the frontend, a Claude-powered chatbot with retrieval-augmented generation answers cited questions, filters the investigation’s content, and visualises supply chain links from fishing vessel to supermarket, handling hundreds to low thousands of user queries per day while the newsroom retains strict editorial control over AI outputs.

AddedOCT 26, 2026
Country index275 tracked
Open implementation ↗Read coverage ↗Visit newsroom ↗
04Audience Engagement
Dimokratiki

Dimokratiki's AI-powered Website

Dimokratiki, a local newspaper in Rhodes, has effectively turned its entire archive and website into an AI‑powered assistant that enhances, rather than replaces, journalists’ work. The system takes reporter-written copy and, under strict “no new facts or judgement” rules, generates concise AI summaries that sit above articles, then converts them into high-quality Greek audio and short vertical videos with synced subtitles, automatically repurposed for podcasts and social platforms. It also processes tens of thousands of archived stories with language models so readers can search more intelligently, discover related coverage and access features like “On this day,” all while offering personalisation tools such as adjustable text, quick reactions, clear labels for AI images and English versions for a multilingual audience.

AddedOCT 26, 2026
Country index11 tracked
Read coverage ↗Visit newsroom ↗
05Journalism
Alaska News

Walter

Alaska News is a reader-funded local news platform that uses an AI reporting agent named “Walter” to automatically turn public records into thousands of local stories. Built by twin founders Lucas and Lee Brown and a small team in Valdez, Alaska, the tool ingests meeting transcripts, filings, notices, and other government documents, then generates articles that are clearly labeled as AI-written and linked directly to their primary sources. A human editor, Cale Green, oversees Walter’s output to keep it accurate and nonpartisan, with the AI focused strictly on “record-based reporting” rather than interviews or on-the-ground work. The broader goal is to address Alaska’s shrinking local news coverage by pairing always-on AI agents with transparent sourcing instead of ad-driven, rage-bait journalism.

AddedOCT 22, 2026
Country index275 tracked
Read coverage ↗Visit newsroom ↗
06Management
WSJ

Doppler News AI

WSJ’s Doppler News AI is an internal analytics tool that ingests and analyzes huge volumes of public online content—tens of millions of TikTok videos and hundreds of thousands of podcast episodes—using embeddings and clustering to detect patterns in what voters are talking about, how they feel, and which issues cut across partisan lines. Instead of predicting election outcomes, it surfaces themes, moods, and anomalies in everyday political conversation that reporters then interrogate, validate, and turn into stories. Each piece built with Doppler involves tight collaboration between engineers, data journalists, and editors, with clear methodological disclosures that emphasize its role as an aid to human judgement, not a replacement for traditional reporting or a fully automated newsroom.

AddedOCT 22, 2026
Country index275 tracked
Read coverage ↗Visit newsroom ↗
07Content Production
Journal Métro

Journal Métro AI-assisted local news relaunch

After Métro Media's 2023 bankruptcy ended more than 30 hyperlocal operations, including Journal Métro and its 16 print weeklies in Montreal and Quebec City, technology entrepreneur Pierre-Antoine Fradet acquired Journal Métro's intellectual property to restore Montreal local coverage with an AI-driven, low-overhead model. The relaunch hired one human editor-in-chief, former Journal Métro journalist Olivier Robichaud. AI tools monitor trusted sources for press releases, reports, and statistics and automatically generate draft articles; Robichaud verifies facts, adds context, seeks opposing viewpoints when needed, and approves publication. Articles are labeled as written by Robichaud or the AI system, with a human in the loop before, during, and after publication.

AddedOCT 22, 2026
Country index10 tracked
Open implementation ↗Read coverage ↗Visit newsroom ↗
08Content Production
Redazione Zero

Redazione Zero autonomous multi-agent newsroom

Redazione Zero is an Italian general-news AI-native outlet whose news selection and writing are left entirely to autonomous multi-agent workflows. Every morning, two agents using Anthropic and OpenAI models independently gather news, which is then filtered through automated checks. The system selects and writes stories according to a public editorial charter, then separate AI verifiers fact-check articles and can block publication; a final AI editor writes an editorial explaining the day's choices. Publisher Emilio Salmeri sets the rules and safeguards but does not select, review, or approve stories before publication, while remaining accountable for corrections and withdrawals. The charter bans crime and court reporting, limits health coverage, protects minors, and generally requires two independent sources.

AddedOCT 21, 2026
Country index18 tracked
Open implementation ↗Read coverage ↗Visit newsroom ↗
09Journalism
Tucson Daily Brief

Tucson Daily Brief AI-native local newsroom

Tucson Daily Brief is a southern Arizona AI-native local outlet founded by technology journalist Nicholas De Leon to expand coverage in a shrinking local media market. It uses a two-tiered publishing model: routine RSS aggregations and simple public directory logs publish automatically, while higher-stakes coverage undergoes manual review. A live AI reporter captures livestream town-hall audio, transcribes it in real time, and drafts articles that De Leon edits and fact-checks before publication (about 90% publication-ready). The project also includes a daily two-minute podcast with an AI clone of De Leon's voice, a Spanish-language TikTok channel with AI anchor Patricia, weekly local crossword puzzles, and ChatTDB, a RAG search engine that answers questions from the Tucson Daily Brief archive.

AddedOCT 19, 2026
Country index275 tracked
Open implementation ↗Read coverage ↗Visit newsroom ↗
10Journalism
Moj Kraj

Moj Kraj AI-native hyperlocal newsroom

Moj Kraj ('My area') is a Croatian AI-native hyperlocal outlet created by AI developer Tihomir Jauk to fill gaps in local news coverage. It covers more than 124 localities by scanning general news plus over 2,000 smaller sources such as municipal and district websites, government announcements, utility providers, traffic updates, and weather information. Content is collected, geotagged, and processed for relevance and reliability. An AI agent pre-screens each source's terms of service and privacy policies for usage and attribution rules before ingestion. Drafts undergo automated grammar, tone, and neutrality checks with numerical thresholds (for example, neutrality below 80% blocks publication), so only about 20–30 of roughly 100 daily drafts go live.

AddedOCT 19, 2026
Country index5 tracked
Open implementation ↗Read coverage ↗Visit newsroom ↗
11Management
El Paso Matters

El Paso Matters workflow

El Paso Matters uses ChatGPT to model campaign pacing, test messaging variations, and run a “reasons for giving” analysis that compares local donors with expatriate supporters. The newsroom inputs donor data into the model, which returns motivation insights that guide segment‑specific outreach and scheduling decisions. Applying the expat‑focused insights, the team crafted a targeted campaign that secured a sizable donation.

AddedOCT 18, 2026
Country index275 tracked
Read coverage ↗Visit newsroom ↗
12Content Production
Hearst

Producer-P

Producer‑P is a Slack‑based AI assistant built by Hearst that uses OpenAI’s GPT‑4 and a fine‑tuned GPT‑3.5‑Turbo model to generate five headline options, SEO‑friendly titles, URLs, related links and notification summaries for journalists’ stories. Reporters invoke the bot in Slack, review and edit the suggestions, and publish the final copy in the CMS, while all users complete a responsible‑AI training that stresses human oversight, transparent summarisation and factual accuracy. Since launch the tool handles more than 1,000 requests per month across the Hearst network, has trained over 200 journalists, and has maintained a clean record of factual errors, prompting further internal AI products for translation, newsletters

AddedOCT 16, 2026
Country index275 tracked
Read coverage ↗Visit newsroom ↗
13Journalism
Connecticut Mirror

CT Mirror AI tools

Connecticut Mirror runs a suite of AI tools and newsroom bots that ingest legislation, meeting transcripts, historical records and state documents, including bots trained on legislative voting records, the state “Blue Book,” and research‑report analyzers. The system scans meetings and public records to surface story leads, summarizes long documents, answers queries about legislators’ voting patterns and helps craft precise FOIA requests, with a planned video‑scraper to turn public video into searchable text. The effort has created an “AI data reporter & product developer” role, expanded coverage to 169 Connecticut cities

AddedOCT 14, 2026
Country index275 tracked
Read coverage ↗Visit newsroom ↗
14Management
Kaktus Media (RU)

Kaktus Media's AI Workflow

Kaktus Media, Kyrgyzstan uses AI selectively across transcription, translation, social-media repurposing, voice generation, and occasional image generation, while keeping strong human editorial control over every output. One of the newsroom’s most useful AI workflows is transcription: journalists use AI to turn long audio or video recordings, particularly lengthy interviews, into text that can then be reviewed and used during reporting and editing. AI is also periodically used for translation between Russian and Kyrgyz, but the resulting translations require substantial human editing because current models are not reliable enough for direct publication. For social-media production, the SMM team uses tools including Gemini and Claude to condense long-form articles into shorter carousel-style posts; the AI-generated summaries and headlines are subsequently edited and proofread before publication.

AddedOCT 14, 2026
Country index2 tracked
Read coverage ↗Visit newsroom ↗
15Journalism
Tobol Info (RU)

Tobol Info AI-assisted Workflow

Tobol Info, Kazakhstan created AI-assisted research and editorial workflows using ChatGPT to deepen reporting and improve formal information requests to government institutions. Journalists first research a topic independently and prepare an initial understanding or draft set of questions. They then use ChatGPT as a second analytical layer to identify gaps, overlooked angles, “blind spots,” and additional questions that may be worth pursuing. The model can also help break a broad issue into subtopics and clarify which government bodies are responsible for different parts of the story.

AddedOCT 14, 2026
Country index3 tracked
Read coverage ↗Visit newsroom ↗
16Management
ArzuNews

ArzuNews Telegram Workflow

ArzuNews, Turkmenistan created AI-powered Telegram workflows that function as a lightweight mobile newsroom, combining news monitoring, editorial coordination, publishing controls, AI translation, audio generation, and AI-assisted development in one interface. The workflow can monitor news sources and surface potentially relevant stories to journalists in Telegram; reporters can take an item into production, after which its status changes to “draft,” while editors can see who is working on it and the system updates the status once the article is published. Editors can also use the Telegram bot to change headlines, assign sections, reject stories, publish immediately, or schedule publication without opening the website CMS. For AI-specific tasks, journalists can send text to Telegram bots that pass it to external AI services for translation or audio generation and return the result directly in the chat.

AddedOCT 14, 2026
Country index1 tracked
Read coverage ↗Visit newsroom ↗
17Content Production
La Voz Chicago

La Voz Chicago AI Translation Workflow

La Voz Chicago, the Spanish‑language outlet of Chicago Public Media, uses the OpenAI API to translate Chicago Sun‑Times articles from English into Spanish with the appropriate tone and dialect. The workflow automatically pulls new Sun‑Times stories, sends them to the API, and places the Spanish draft—including photo captions—in a Google Doc where editors review, edit, and publish, adding a note that the piece was AI‑translated. The system has cut publishing lag from two days to same‑day, as illustrated when a profile of the new Pope was posted in Spanish within hours and generated five times the normal traffic.

AddedOCT 11, 2026
Country index275 tracked
Read coverage ↗Visit newsroom ↗
18Journalism
Alma Media

Kielikaveri

Kielikaveri (“Language Buddy”) is an in‑house editorial AI assistant used across Alma Media’s Finnish newsrooms to improve language quality and streamline writing. It’s available both as a browser tool and directly inside the content management system, where it highlights suggested edits, explains them based on each newsroom’s own language guidelines, and lets journalists accept or reject changes with one click. The tool focuses on clarity, typos, and nuanced Finnish usage, and is continuously refined through direct journalist feedback via the CMS and a dedicated Slack channel, as well as regular face‑to‑face tuning sessions. Built and redesigned by Alma’s AI team to be cost‑effective and tailored to Finnish (a relatively small language globally), Kielikaveri has evolved with new language models and now sits among roughly forty AI tools that are widely used in Alma’s editorial workflows.

AddedOCT 7, 2026
Country index17 tracked
Read coverage ↗Visit newsroom ↗
19Distribution
Lisa AI

Lisa AI

Lisa AI is a free Argentinian news app that uses artificial intelligence to generate personalized, 10‑minute audio briefings based on each user’s chosen topics, pulling from dozens of local news outlets and clearly attributing every summarized story to the original publishers. Users set their interests, Lisa continuously scans recent coverage (excluding paywalled content), and then assembles an AI-voiced newscast in an authentic Argentinian accent, with an in‑app chat that links back to the full articles. Built by two 23‑year‑old founders and a small volunteer team, the app aims to rebuild trust in professional journalism by matching young audiences’ preferred formats (podcasts, feeds, chatbots) while preserving editorial curation and source transparency instead of replacing newsrooms.

AddedOCT 2, 2026
Country index37 tracked
Open implementation ↗Read coverage ↗Visit newsroom ↗
20Content Production
The Visa Pulse

The Visa Pulse: AI-powered immigration policy coverage

The Visa Pulse is an independent publication using AI to turn U.S. immigration policy changes into plain-language coverage for international students, workers and other visa holders. Its pipeline retrieves government documents and selected reporting, filters for relevance, groups related coverage and generates articles explaining what changed, who is affected and which dates matter. Anthropic Claude models support relevance screening and article generation. Coverage includes visa-type filters, impact labels and source links. Articles can publish automatically; the site distinguishes automated checks from recorded human review and provides a route for corrections. Its coverage is informational, not legal advice. The project demonstrates how a small specialist publication can use AI across discovery and content production while making its sourcing and review limits visible.

AddedOCT 1, 2026
Country index275 tracked
Open implementation ↗Read coverage ↗Visit newsroom ↗
21Journalism
The New York Times

The Epstein Files Engine

The Epstein Files Engine is an AI-powered investigative tool The New York Times built to help reporters search, connect, and verify information across a massive trove of Jeffrey Epstein–related documents, the Times’s own archive, and external news coverage. Reporters type questions in natural language, and the system uses a large language model to plan and generate Google BigQuery SQL queries over three corpora, then returns answers with strong citations so journalists can audit every claim against primary material. A key component is Diff, a text-and-visual duplicate detection method that filters out redundant documents and highlights genuinely novel findings, allowing reporters to focus on new information instead of re-reading near-duplicates.

AddedSEP 30, 2026
Country index275 tracked
Open implementation ↗Read coverage ↗Visit newsroom ↗

Showing 21 of 1359 in this snapshot · 1359 total in the live archive.

Contribute

Know a project we missed?

Add a newsroom AI project, product or experiment. Every submission is reviewed before it joins the index.

For AI assistants

Query the full index via MCP.

Search, compare and analyze implementations directly from your preferred AI assistant.

Explore our free MCP →