The crisis in trust
The World Economic Forum’s 2025 Global Risks Report identifies misinformation and disinformation as the world’s leading risk for 2027, describing “a proliferation of false or misleading content complicating the geopolitical environment.”
Trust in journalism has languished at 40% globally for three consecutive years, following years of decline. In the UK and Germany, confidence has dropped 16 and 15 points respectively since 2015.
Today, 58% of people worldwide cannot distinguish real from fake news online – rising to 73% in the United States. (Reuters Institute 2025 Digital News Report.)
The AI Acceleration
AI has transformed disinformation from a manageable challenge into a systemic threat. Generative AI now produces millions of credible-seeming articles, deepfake videos, and synthetic images at near-zero marginal cost.
What once required state-level resources is now available to any bad actor through consumer applications. Defence chronically lags attack in this asymmetric arms race.
The AI Paradox
AI detection tools remain unreliable as primary defences against synthetic content. Even sophisticated systems produce high rates of false positives when analysing human-created content.
AI detection also faces a fundamental asymmetry. As generation techniques evolve, detection systems lag perpetually behind the increasing volume, speed and sophistication of disinformation.
The Commercial Fallout
Only 18% of consumers in wealthy countries pay for online news. News avoidance has reached 40% – the highest ever recorded. Quality journalism struggles to command premium revenues from both subscribers and advertisers
Audiences are migrating to unaccountable personality-driven channels on social media, which overtakes traditional news sources in the US for the first time.
A Way Forward
The Reuters Institute 2025 data shows that “trusted news brands remain the first choice for many when it comes to checking information, or alerting them to important breaking news, even if people don’t need them as often as they once did.”
But traditional fact-checking alone cannot resolve these threats. Verification must be embedded into content itself, and made tamper-proof from creation through to publication.