A healthy information environment depends on the supply and demand of quality information. To remain competitive, news outlets, publishers, research institutes and NGOs must develop tools to verify where information comes from and whether it can be trusted. Otherwise, they risk losing the independence needed to serve the public interest.
In September, unredacted files from The New York Times’ lawsuit against OpenAI and Microsoft revealed that one of Microsoft’s senior scientists had described the use of copyrighted material to train AI models as the “largest theft of labor in human history”.
Publishers are pushing back. According to NewsGuard, 67 per cent of the most-engaged news sites now block crawlers from accessing their archives, while lawsuits are mounting. But blocking is defensive, leaving a question unanswered: what could publishers do with those archives themselves?
A news outlet’s archive is an asset AI companies cannot replicate. In Guatemala and South Africa, media organisations and nonprofits are using archives to build chatbots that answer farmers’ questions, tools that help readers examine procurement databases, and systems that sort reader comments.
Publishers’ influence over information markets will depend on their ability to innovate. Big Tech, however, continues to set the incentives. While publishers take large language model developers to court, they are also trying to track where their content appears and charge bots and AI agents for access.
Organisations with archives are already using AI to gain advantages once beyond their reach. With information more plentiful than ever, news outlets are seeking new ways to build trust and reach audiences. Some are selling real-time data and actionable analysis, while others are exploring crowd-powered prediction markets.
Independent publishers are developing their own AI tools, running models locally and mapping their data architecture. They are training models on verifiable proprietary material and exploring community-based, open-source and commercial applications from the earliest stages of product development.
These investments can create revenue through native advertising, sponsorships and in-house AI products. Food for Mzansi, an agriculture-focused publisher in South Africa, has developed Agri-Pulse 360, an AI ecosystem offering small-scale commercial farmers round-the-clock advice. Drawing on an eight-year archive of thousands of articles, its chatbot uses retrieval-augmented generation (RAG) to answer farmers’ questions. It is expected to become available through WhatsApp. It can also generate TikTok reels and track newsletter reading habits.
Food for Mzansi hopes to attract advertisers through a native advertising tool called “Solution Moments”. It is seeking brand sponsorships for AI-generated workflows and plans to automate promotions in targeted newsletters.
In Guatemala, independent news outlet Ojoconmipisto and data provider Sociedad.info have launched Pin Local, an AI tool that uses government spending records to produce personalised weekly newsletters. Ojoconmipisto journalists provide the reporting, while Sociedad.info’s AI searches its procurement database for information relevant to readers in Guatemala’s 22 administrative regions. The system could use data from other countries and support the sale of customised reports on individual contracts.
In Argentina, fact-checking organisation Chequeado uses AI to analyse thousands of reader comments. Its tool, Nunchi, classifies comments by sentiment and intent, then ranks them by relevance. Trained to recognise humour and irony, it sends editors twice-daily Slack briefs highlighting recurring topics and unanswered questions. A dashboard tracks changes in audience interests.
In Colombia, Cuestión Pública has built six Model Context Protocol (MCP) servers connecting 13 public procurement databases, a national business register and its reporting archive. Users can explore records on nine million companies, political events and politically connected individuals in a single conversation. It tracks token use live and is developing an accountability-reporting tool, with plans for a paid, privacy-focused version.
These projects reflect a search for revenue beyond traditional news products. Some publishers hope to build loyalty through direct relationships with readers; others see information as a commodity to sell to AI firms by the unit.
Either way, publishers must understand how their content is used and retain control over it. To avoid becoming mere content suppliers for Big Tech, they need the freedom and resources to experiment, develop new products and scale them on their own terms.
The writer is Director of Media Innovation at the International Press Institute.




































