A Society That Cannot Tell Real From Fake Can Be Steered

Journalism in the age of AI: The New Rules of Verification & trustful Dissemination

An image from Iran appears on social media. No Danish photographers are in the country. Within two hours it has been shared tens of thousands of times. A newsroom has to decide: do we publish it or not, and what do we tell the audience about what we do not know?

This is the daily reality of a newsroom in 2026. It is also a security problem. Information is the first domain in which conflicts are now fought, and a population that cannot tell real from fabricated is easier to influence. The share of Danes concerned about misinformation rose from 43 to 49 percent in a single year.

The technical problem is familiar from information security. Provenance depends on signatures, watermarks and trust lists. Those signals can be stripped, spoofed or simply ignored. The EU AI Act has required deepfakes to be labelled, but a label solves only half the problem: a signal the audience does not understand, or does not notice, has no effect.

No single actor can fix this. Verification is an ecosystem running from the camera, through the watermark and signature that follow the file, through the platforms that strip or preserve that information, to the newsroom that assesses it, and finally to the citizen who has to make sense of it. It is critical infrastructure with no owner, and the chain only holds if every link participates. Today, several do not.

Three of those parties meet here:

The newsroom: Marie Buss, Senior Editorial Manager at DR News, representing DR Verificerer, on the operational reality of verification and on the hardest part: telling the audience what has been verified and what has not.

The society: Mads Vestergaard, philosopher and researcher at Roskilde University, on the democratic cost, and on how doubt about images is used strategically to dismiss what is real.

The technology: Jeppe Nørregaard, founder of InReality, with the IT University of Copenhagen, on cryptographic provenance for images and video, what it can and cannot prove, and the unsolved step of making it understandable to a general audience.

Moderated by Jari Kickbusch, IT University of Copenhagen.