Artificial intelligence has radically changed the economics of information.
AI systems can produce, replicate and target content at unprecedented speed and scale. In the case of fake news, this creates an important asymmetry: the cost of producing convincing deceptive content can fall while the cost of verifying that information rises.
This is the central problem explored by Cosmin Marinescu and Prof. Dr. Karina Ochiș in their newly published academic article:
“Fake News, Artificial Intelligence, and Transaction Costs: From Misinformation to Regulation.”
Published in Amfiteatru Economic, the study applies transaction cost economics to AI-mediated information markets and introduces a distinction between two forms of cost:
- Market Transaction Costs (MTCs): costs associated with activities such as producing, coordinating, searching for and verifying information.
- Imposed Transaction Costs (ITCs): costs created through regulation, including documentation, auditing, compliance and administrative oversight.
The research argues that AI does not simply reduce transaction costs.
Instead, AI redistributes them.
For actors producing deceptive information, AI can dramatically lower production and coordination costs. For journalists, platforms, researchers, organizations and ordinary users attempting to establish what is credible, verification and evaluation become increasingly demanding.
Regulation then changes the equation again.
Three approaches to AI governance
The article compares the regulatory architectures of the United States, China and the European Union.
The United States represents a comparatively market-led model, with relatively low domestic compliance burdens and greater regulatory concentration in security-sensitive areas.
China follows a more state-led and centralized model, where regulation and industrial policy operate closely together and information governance aligns strongly with state objectives.
The European Union takes a legalistic, rights-based approach, using instruments such as the AI Act and GDPR to introduce broad risk-based obligations and safeguards.
Each system therefore creates a different distribution of costs for platforms, businesses, public-interest organizations and users.
Rethinking what effective AI regulation means
The study’s central implication is significant.
The effectiveness of AI governance should not be evaluated solely according to the intensity of regulation.
The more important question is:
How are the costs created by AI and AI regulation distributed across the actors operating within the information ecosystem?
Regulation intended to limit harmful uses of AI may also create substantial burdens for smaller organizations, journalists, civic projects and innovators attempting to use AI responsibly.
Understanding those trade-offs becomes increasingly important as governments develop the next generation of AI governance frameworks.
Publication details
Marinescu, C. and Ochiș, K. (2026).
Fake News, Artificial Intelligence, and Transaction Costs: From Misinformation to Regulation.
Amfiteatru Economic, 28(73), pp. 1114–1130.
DOI: 10.24818/EA/2026/73/1114
