AI, Misinformation & Political Views: How Online Images Shape Beliefs

As artificial intelligence technologies advance, concerns are mounting regarding their capacity to distort political discourse and erode public trust in media. On April 2, 2026, Connecticut Public’s program The Wheelhouse examined how AI-generated content is influencing perception during a period when trust in news organizations has reached historic lows. The discussion highlighted specific recent events where digital imagery and automated narratives may have altered public understanding of critical incidents.

During the broadcast, hosts pointed to high-profile cases where online content potentially obscured factual realities. References were made to coverage surrounding the fatal shooting of a Minneapolis woman by federal law enforcement, as well as reports describing conflict in Iran. National Public Radio characterized the situation in Iran as America’s first AI-fueled war in reporting dated March 26, 2026. These examples illustrate the growing complexity journalists and audiences face when distinguishing between verified events and synthesized media.

The erosion of confidence in traditional information sources provides a backdrop for these technological challenges. Links shared in connection with the program cited Gallup polling indicating that trust in media has reached new lows. This sentiment aligns with academic findings regarding the impact of generative AI on information consumption. A scoping review published in the journal AI & SOCIETY on September 30, 2025, analyzed 24 empirical studies and concluded that exposure to AI-generated misinformation reduces trust and influences decision-making.

Research on Image and Text Manipulation

Academic institutions are increasingly documenting the mechanisms by which AI tools affect belief systems. Researchers at Harvard University conducted a pre-registered experiment examining how properties of AI-synthesized images influence belief in misinformation. The study found that realistic and probative images, defined as those providing strong evidence to headlines, predicted greater belief in false headlines. This suggests that the visual fidelity of generated content plays a significant role in its persuasive power.

Beyond static imagery, large language models are also implicated in the spread of false claims. The scoping review published in AI & SOCIETY noted that large language models can generate highly convincing misinformation. These systems often exploit cognitive biases and ideological leanings of audiences. While some capabilities exist to detect false claims and enhance user resistance, mitigation efforts have shown mixed results. The review emphasized that safeguards are inconsistently applied across different platforms.

Autonomous Propaganda Capabilities

The scale of potential disinformation campaigns may be expanding beyond human coordination. A study from the USC Viterbi School of Engineering found that AI agents can autonomously coordinate propaganda campaigns without human direction. This development indicates that malicious information operations could potentially scale without direct human oversight, complicating efforts to trace the origin of coordinated narratives.

Looking toward the immediate future, the World Economic Forum addressed these risks in a report published on March 12, 2026. The organization outlined how cognitive manipulation and AI will shape disinformation in 2026. The report noted that technologies such as social media, online chat, and artificial intelligence are allowing opportunistic actors to target audiences with greater precision. These tools enable the rapid creation and dissemination of content designed to exploit psychological vulnerabilities.

Policy and Election Implications

The intersection of generative AI and democratic processes remains a key area of scrutiny. Knight Columbia published an assessment titled Don’t Panic (Yet): Assessing the Evidence and Discourse Around Generative AI, and Elections. This analysis contributes to the broader discourse on how these technologies intersect with electoral integrity. Within the Connecticut Public broadcast, federal policy reporter Lisa Hagen participated in the discussion, bringing a focus on regulatory and governmental responses to these challenges.

Academic experts joined the conversation to provide context on media literacy and communication studies. Alyssa Appelman, an associate professor in the William White School of Journalism and Mass Communications at Kansas University, contributed to the program. Renee Hobbs, a professor in the Department of Communication Studies at the University of Rhode Island, also participated as a guest. Their involvement underscores the interdisciplinary effort required to address the dual role of generative AI in both generating and detecting misinformation.

The scoping review in AI & SOCIETY underscored the need for standardized evaluation metrics and stronger regulatory measures. The authors called for interdisciplinary collaboration to ensure the responsible use of generative AI in the information ecosystem. As personalized corrections prove effective in some contexts, the inconsistency of safeguards remains a hurdle for widespread mitigation. The review highlighted that while opportunities exist to fight misinformation, the challenges introduced by rapid advancements are significant.

The Wheelhouse episode is available for public access across multiple audio platforms. The program can be accessed via Apple Podcasts, Spotify, Google Podcasts, Amazon Music, TuneIn, and Listen Notes. Connecticut Public noted that contributors Megan Rodriguez-Hawkins and Divina Cordeiro assisted with the show. The discussion serves as part of an ongoing examination of how technology influences public perception during a period of heightened skepticism toward media institutions.