Disinformation

Definition

Disinformation refers to deliberately fabricated or manipulated information disseminated with the intent to deceive, influence public opinion, or cause societal harm. It is distinct from misinformation, which involves the unintentional spread of inaccurate content. In the context of AI-enabled threats, disinformation encompasses the use of generative AI tools to produce synthetic text, images, audio, and video at scale, making false narratives more convincing and harder to detect. Disinformation campaigns may target political processes, public health discourse, financial markets, or social cohesion. The deliberate nature of disinformation distinguishes it as a strategic instrument rather than an incidental by-product of information systems.

How It Relates to AI Threats

Disinformation intersects with AI threats primarily within the Information Integrity domain. Generative AI has dramatically lowered the cost and increased the speed of producing convincing false content, including synthetic media and fabricated news articles. AI-powered disinformation campaigns can target specific demographics with personalised false narratives, erode trust in legitimate institutions, and manipulate democratic processes. When combined with deepfakes, AI-generated disinformation becomes particularly difficult for both humans and automated systems to distinguish from authentic content.

Why It Occurs

• Generative AI tools enable rapid production of realistic text, audio, and visual content at minimal cost
• Social media platforms amplify false content through algorithmic recommendation systems
• Attribution of AI-generated content is technically difficult, reducing accountability
• State and non-state actors have strategic incentives to manipulate public discourse
• Detection and verification mechanisms have not kept pace with generation capabilities

Real-World Context

The 2023 Slovakia parliamentary election demonstrated the impact of AI-enabled disinformation when a fabricated audio recording, purporting to capture a candidate discussing vote-buying, circulated widely in the final days before voting. The recording was produced using AI voice synthesis and proved difficult for fact-checkers to debunk within the compressed pre-election timeline. This incident illustrates how AI-generated disinformation can be timed to exploit the gap between content creation and verification capacity.