False or misleading information, called misinformation when it is shared in good faith and disinformation when it is spread deliberately to deceive. AI matters here because it makes convincing text, images, audio and video cheap to produce in any language and at any volume. How much that changes what people believe is disputed: some studies of recent elections found little measurable effect, while others warn the deeper damage is to trust in genuine evidence.
Technical view
The distinction usually follows Wardle and Derakhshan’s 2017 Council of Europe report on ‘information disorder’, which adds malinformation: genuine information shared to cause harm. Generative models lower the cost of persuasive content, personalised messaging and fake personas, and can produce misinformation themselves through hallucination. Countermeasures include provenance and watermarking, platform moderation, detection classifiers and media literacy, each with gaps: detectors are unreliable, watermarks can be stripped, and widespread fakery lets people dismiss authentic material as fake. Researchers at the Alan Turing Institute found no evidence that AI-generated content meaningfully affected the results of the 2024 United Kingdom, French or European Parliament elections; whether that holds as tools improve is debated.
Related
Part of AI Terminology Index, section: Problems, Risks and Ambitions.