DistantNews
Support us
🇬🇷 Greece /Technology

Invisible Intervention: How AI is Secretly Changing Our Posts and Steering Public Opinion

From Ta Nea · () Greek

Translated from Greek, summarized and contextualized by DistantNews.

At a glance

Analysis Documents & data New plan
  • AI tools used for social media posts can introduce biases and influence public opinion, according to a study by Oxford Internet Institute and Hasso Plattner Institute.
  • Large language models systematically alter the stance of posts on controversial issues, even when instructed to maintain the original meaning.
  • The findings raise concerns about AI's growing role in shaping public discourse and the need for transparency and accountability.

Artificial intelligence tools used to create or edit social media posts can subtly embed biases that spread online and sway public opinion, a new study reveals. Research from the Oxford Internet Institute (OII) at the University of Oxford and the Hasso Plattner Institute at the University of Potsdam found that large language models (LLMs) systematically change the stance of posts on contentious topics.

The tools of artificial intelligence used for writing or editing texts can introduce biases into users' posts.

— Study AuthorsDescribing a key finding about AI's impact on social media content.

Even when instructed to preserve the original meaning, these AI models were observed to alter the tone and perspective of the content. Through simulations based on real social network data, researchers discovered that small modifications can accumulate over time, gradually influencing how collective opinion is formed. This suggests that AI-mediated communication could become a powerful new mechanism for shaping public discourse.

The study, titled "AI-mediated communication can influence collective opinion," highlights concerns about the increasing use of AI writing tools on social media. The researchers identified similar patterns of bias across different AI systems. Several models were found to reinforce specific viewpoints on issues like gun control, cannabis legalization, and feminism, while adopting a more negative stance on topics such as atheism and the death penalty.

Small changes in individual posts can affect public opinion over time.

— Study AuthorsExplaining the cumulative effect of AI-driven content modifications.

These findings raise significant questions about transparency and accountability in the digital public sphere. The research indicates that the source of bias is not solely within the AI models themselves; the design choices made by platforms when developing these tools also play a crucial role in determining the extent of AI's influence. The study will be presented at upcoming AI4Good and Technical AI Governance Research workshops at the International Conference on Machine Learning (ICML 2026) in Seoul.

AI-mediated communication creates a new way of influencing public discourse, raising issues of transparency and accountability.

— Study AuthorsHighlighting the implications of AI in shaping public dialogue.
DistantNews Editorial

Originally published by Ta Nea in Greek. Translated, summarized, and contextualized by our editorial team with added local perspective. Read our editorial standards.

More Stories

From Our Blog