Open LLMs Tested for Online Political Influence
A new study examines how openly accessible large language models (LLMs) can be utilised to conduct online political influence campaigns. The research focuses on the models' ability to express political opinions.
What happened?
Researchers have developed a framework to test how effectively open LLMs can be used to spread political influence online. The study, published on arXiv on 26 May 2026, analysed over 30 LLMs from ten model families and five countries of origin. The focus was on locally deployed models, as these are considered more relevant for actors prioritising anonymity. The framework measured the models' 'Overton Windows' (OWs) — the range of political opinions a model can express — and how 'jailbreaks' affect this range.
Key facts
| Publikationsdatum | 26 maj 2026 |
|---|---|
| Antal modeller analyserade | Över 30 |
| Antal modellfamiljer | 10 |
| Antal ursprungsländer | 5 |
| Koncept | Overton Windows (OWs) |
”As large language model (LLM)-based agents increasingly participate in online discourse, red-teaming their capacity to support political influence campaigns is critical for information integrity.”
”We find systematic asymmetries in political expressivity: open-source LLMs are typically more willing to generate left-leaning social media content.”
Why it matters
The study highlights a critical aspect of information integrity: the increasing risk of LLM-based agents being used in political influence campaigns. By measuring 'Overton Windows', the models' political expressiveness is quantified, revealing asymmetries in content generation capabilities. The results show that many open LLMs more frequently generate left-leaning content on social media. This insight is crucial for understanding the potential for manipulation of public opinion formation.
Who is affected?
Researchers in AI and information security are directly affected by insights into LLM vulnerabilities. Developers of open LLMs receive vital feedback regarding potential misuse. Social media users may be indirectly affected, as the study addresses methods for creating misleading political content. Authorities and regulatory bodies working with information integrity gain new data to consider.
Impact on the EU
According to the study, open LLMs are a global issue, featuring models from various countries of origin. EU citizens may therefore be indirectly affected by information generated by these models. The issue of AI misuse for political influence is relevant to potential future AI legislation within the EU, such as the AI Act.
What else you should know
The study focused specifically on 'locally deployed open-source LLMs', distinguishing it from research on API-based models. it emphasises how anonymity considerations influence the choice of technology for malicious actors.
Quick answers about this story
Vad har hänt?
När hände det?
Varför spelar det roll?
Vilka bolag berörs?
The link opens in a new window and leads to the publisher's own site.
Källan har spårats automatiskt från utgivaren via Aheadlines signalkedja.
Get similar news straight to your inbox
The reader's room
Send in a question or an addition. The newsroom reads everything before it's published and replies when relevant. No AI-generated text – just people.
Sign in to submit a comment or question.
Read the article through your role
- Decide whether this affects strategy over 6–12 months or is just noise.
- Discuss with leadership: do we own the right question or does ownership need to move?
- Ask: what risk are we taking by NOT acting on this this quarter?
Generated angle — not editorial analysis of "Open LLMs Tested for Online Political Influence"