Does AI-generated misinformation influence elections?

Numbers match the citations in the article. A working link isn't proof that a page supports a claim; check the quoted passage and date.

1. 1

   Large language models can consistently generate high-quality content for election disinformation operations.

   PloS one (Williams et al.)Published Mar 17, 2025Checked Oct 4, 2026

   “Using DisElect, we test 13 LLMs and find that most models broadly comply with these requests; we also find that the few models which refuse malicious prompts also refuse benign election-related prompts, and are more likely to refuse to generate content from a right-wing perspective. Secondly, we conduct a series of experiments (N = 2 , 340) to assess the "humanness" of LLMs: the extent to which disinformation operation content generated by an LLM is able to pass as human-written. Our experiments suggest that almost all LLMs tested released since 2022 produce election disinformation operation content indiscernible by human evaluators over 50% of the time. Notably, we observe that multiple models achieve above-human levels of humanness. Taken together, these findings suggest that current LLMs can be used to generate high-quality content for election disinformation operations, even in hyperlocalised scenarios, at far lower costs than traditional methods, and offer researchers and policymakers an empirical benchmark for the measurement and evaluation of these capabilities in current and future models.”

2. 2

   An AI-driven conceptual framework for detecting fake news and deepfake content: a systematic review.

   Frontiers in artificial intelligence (Moyo et al.)Published Mar 2, 2026Checked Oct 4, 2026

   “From an initial set of 120 database records, complemented by citation chaining, 34 studies published between 2014 and 2025 were included for analysis. Eighteen studies focused on deepfake generation and detection models, eight examined social and behavioural implications, and eight addressed ethical and regulatory frameworks. Thematic synthesis reveals a clear methodological shift from convolutional neural networks toward transformer- and CLIP-based architectures, alongside the emergence of large-scale benchmark datasets. However, persistent challenges remain in multimodal detection, cross-dataset generalization, explainability-robustness trade-offs, and the translation of governance principles into deployable systems. This review contributes an integrated conceptual framework that operationally connects detection technologies, explainable AI (XAI), and governance mechanisms through explicit feedback loops. Future research directions emphasize robust multimodal benchmarks, retrieval-augmented detection systems, and interdisciplinary approaches that align technical innovation with ethical and policy safeguards.”

3. 3

   AI-driven disinformation: policy recommendations for democratic resilience.

   Frontiers in artificial intelligence (Romanishyn et al.)Published Jul 31, 2025Checked Oct 4, 2026

   “The regulatory asymmetry between traditional media - historically subject to public oversight - and digital platforms exacerbates these vulnerabilities. This policy and practice review has three primary aims: (1) to document and analyze the role of AI in recent disinformation campaigns, (2) to assess the effectiveness and limitations of existing AI governance frameworks in mitigating disinformation risks, and (3) to formulate evidence-informed policy recommendations to strengthen institutional resilience. Drawing on qualitative analysis of case studies and regulatory trends, we argue for the urgent need to embed AI-specific oversight mechanisms within democratic governance systems. We recommend a multi-stakeholder approach involving platform accountability, enforceable regulatory harmonization across jurisdictions, and sustained civic education to foster digital literacy and cognitive resilience as defenses against malign information. Without such interventions, democratic processes risk becoming increasingly susceptible to manipulation, delegitimization, and systemic erosion.”

4. 4

   Citizens Versus the Internet: Confronting Digital Challenges With Cognitive Tools.

   Psychological science in the public interest : a journal of the American Psychological Society (Kozyreva et al.)Published Dec 1, 2020Checked Oct 4, 2026

   “We begin by systematically outlining how online and offline environments differ despite being increasingly inextricable. We then identify four major types of challenges that users encounter in online environments: persuasive and manipulative choice architectures, AI-assisted information architectures, false and misleading information, and distracting environments. Next, we turn to how psychological science can inform interventions to counteract these challenges of the digital world. After distinguishing among three types of behavioral and cognitive interventions-nudges, technocognition, and boosts-we focus on boosts, of which we identify two main groups: (a) those aimed at enhancing people's agency in their digital environments (e.g., self-nudging, deliberate ignorance) and (b) those aimed at boosting competencies of reasoning and resilience to manipulation (e.g., simple decision aids, inoculation). These cognitive tools are designed to foster the civility of online discourse and protect reason and human autonomy against manipulative choice architectures, attention-grabbing techniques, and the spread of false information.”

5. 5

   Misinformation (Wikipedia)

   WikipediaPublished Oct 3, 2026Checked Oct 4, 2026

   “Misinformation is incorrect or misleading information. Whereas misinformation can exist with or without specific malicious intent, disinformation is deliberately deceptive and intentionally propagated. Misinformation is typically spread unintentionally, mostly caused by a lack of knowledge, an error, or simply a misunderstanding, which contrasts with disinformation. Misinformation can include inaccurate, incomplete, misleading, or false information as well as selective or half-truths. Social media platforms, such as Facebook, Instagram, X, etc., are designed in ways that enable information, including misinformation, to be posted and shared far more quickly than through other communication mediums. In January 2024, the World Economic Forum identified misinformation and disinformation, propagated by both internal and external interests, to "widen societal and political divides" as the most severe global risks in the short term. The reason is that misinformation can influence people's beliefs about communities, politics, medicine, and more.”