Aspen Policy Academy

REUTERS/Ahmed Saad

AI is undermining OSINT’s core assumptions. Here’s how journalists should adapt

  • Article Published December 12, 2025

This article originally appeared on the Reuters Institute website on December 12, 2025.

By Georgia Edwards, Zuzanna Wojciak, and Influencing AI from the Outside alum Shirin Alren

The fast evolution of generative AI makes it increasingly easy to create visual content that looks indistinguishably real. As AI-generated content mimics authentic scenes and locations, open-source investigators face a new challenge: verifying what’s “real” in a world where reality itself can be fabricated. 

With the release of advanced video generation models like Veo 3 and Sora 2, generative systems can now create entire moving sequences from text, images, or other audiovisual inputs. As these models advance, the boundary between authentic footage and AI-generated fabrication will become more difficult to discern. 

Browse Related Articles

Green code against a black screen.

In Pentagon-Anthropic standoff, AI is real-time testing the balance of power in future of warfare

This article originally appeared on CNBC on February 27, 2026.
Mock code for an AI Large Language Model (LLM) that could intelligently answer questions.

How to Manage Misinformation in Large Language Models

This article originally appeared on Tech Policy Press on February 25, 2026.
Ring surveillance camera at a front door.

The Spy Next Door: Are Smart Doorbells Building a Surveillance State?

This article originally appeared on SC Media on January 28, 2026.