Most conversations about AI and the news begin with pessimism. But a series of summits seeded by IHS began with a hopeful mindset that we can use AI to build better institutions, and a year later the ideas are showing up in newsrooms.
Within a few years, most people won't read articles from legacy newspapers and traditional news sites. They'll ask an AI assistant to discover, analyze, and organize a boutique news feed tailored specifically to their interests. The decline of legacy news outlets usually arrives wrapped in dread about the end of a shared reality and an endless flood of machine-made slop. Those worries aren't unfounded. But they aren't the only way to look at the question, and over the past year a small group of scholars, builders, and investors gathered to ask other questions: What do we actually want from a fully AI-mediated information ecosystem? How could AI be deployed instead to widen everyone's access to accurate, relevant information in ways we never could have when the news depended entirely on merely human production? What new kinds of novel information products might we build, if we moved past the incremental and the fearful and instead let ourselves imagine?
These are not idle questions. When people stop trusting the news, the press can no longer do its job of holding power accountable, and that distrust seeps into the other institutions a free society leans on. Getting the next information ecosystem right is the kind of problem IHS built its Trust and Governance Initiative (TGI) to take on; we find and back the scholars asking the right questions.
In early 2026, that meant a bet on Shuwei Fang, a fellow at Harvard's Shorenstein Center. Fang was studying how an AI-mediated internet could function differently from the attention-economy we have become all too familiar with.
That seed paid for a series of working summits called Signals at Scale, held in New York, San Francisco, Cambridge, and most recently in Copenhagen. Each brought together carefully chosen thinkers, builders, operators, and investors with real track records at the frontier of AI and news.
These programs were invitation-only and deliberately unpublicized, and they were structured to produce a working map of the emerging AI-mediated information ecosystem as well as a clear description of tangible investment opportunities. The conversations, in the words of several participants, "Began where other discussions of AI-mediated information end."
"These summits would not have been possible without the support of IHS." – Shuwei Fang
IHS did more than write the first check. Its early commitment brought in more than twice as much funding from other funders, including the International Fund for Public Interest Media, the Soros Economic Development Fund, and the Financial Times. The relationship was anchored at IHS with Paul Matzko, who leads the Trust and Governance Initiative, and Stephen Wong. IHS supplied much of the administrative backbone for the program, working alongside co-sponsors such as Harvard's Shorenstein Center, and sourced many of the academics who took part. "IHS was the first to commit funding for the series, enabling us to raise further funding," Fang says. "These summits would not have been possible without the support of IHS."
The results outran the plan. Despite being unpublicized, the summits drew enough demand that the original plan for a single gathering grew into four meetings in New York, San Francisco, Cambridge, and Copenhagen; they happened because participants asked for more. The cohorts have stayed engaged, already spinning off collaborations and projects. Out of the series, Fang is now co-founding a venture studio to back startups building this new infrastructure. And in June, Fang was named a senior fellow at CUNY's Craig Newmark Graduate School of Journalism.
And the ideas have begun to achieve escape velocity. For instance, after Gard Steiro—who runs VG, Norway's largest news outlet and part of the Schibsted media group—attended a Signals at Scale program, he turned Fang's argument about "liquid content" (in an AI-mediated world, the value of any single fixed article falls toward zero), into a series of experiments at VG. Rather than publish a piece once, leaving readers to chase later developments on their own, VG’s newsroom is testing stories that stay alive. Each user gets their own boutique running story which AI assembles on the fly to catch a reader up on whatever has changed since they last looked. "This went from an open question to a working newsroom in about a year," Matzko says. "And it's a market answer to the decline of legacy news, built by people who saw an opening rather than waited for a subsidy."
For Matzko, that arc is the whole IHS model in miniature: spot an opening, invest seed money, use it to draw in more, and connect IHS’s network of top scholars to the people who can build on their work.
The news media is only one front in the Trust and Governance Initiative. Through the Estonia Lab, IHS is studying how a country that runs most of its government systems online can model better practices in America. And with the Federal Communications Commission as a test case, IHS is convening scholars and former commissioners to ask what an agency built for the age of radio should become in an age of AI and decentralized media, or whether it still needs to exist at all.
"The ideas need to be lying around before we need them." – Paul Matzko
The overall aim by 2028 is a working shelf of reform blueprints—all built from the conviction that authority works better when it's limited and accountable—that will be ready for when the next crisis pries a stale institution open and creates an opportunity for reform. "The ideas need to be lying around before we need them," Matzko says. "My job is to make sure the right ones are."
The Institute for Humane Studies is America's fastest-growing nonpartisan network grounded in the classical liberal tradition, advancing human freedom and flourishing through evidence-based ideas, research, and a shared belief that every person deserves the opportunity to build a life of meaning, purpose, and fulfillment.