Innocuous-Seeming Data, Latent Ideology: Ideological Generalisation in Finetuned LLMs
arXiv:2607.14888v1 Announce Type: cross Abstract: Finetuning language models on small, curated datasets is standard practice for adapting them to specific policies or domains. We show that finetuning on narrow, factually-defensible, moderation-passing data can cause broad ideological shifts across unrelated domains, while preserving general capabilities. Training GPT-4.1 on right- or left-leaning economics Q&A yields matched ideological shifts on topics such as criminal justice, the environment,
Record details
Published: 17 July 2026
Source: arXiv cs.CY
Category: Research
Topics: Environment
Retrieved: 17 July 2026
Related evidence
These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.
Climate-control ChatGPT for EFL writing and AI literacy
Frontiers in Artificial Intelligence · 22 July 2026
OpenAI single-agent LLM architecture reduces computational overhead relative to multi-agent orchestration in a simulated mars rover decision-support benchmark
Frontiers in Robotics and AI · 6 July 2026
Anthropomorphizing AI: Perceptions, Interpretations, and the Humanization of Generative AI
Social Media + Society · 3 August 2026
A Counterexample to Fourier Alignment in Single-Neuron Modular Addition
arXiv cs.LG · 5 August 2026
A Counterexample to Fourier Alignment in Single-Neuron Modular Addition
arXiv cs.LG · 5 August 2026
Fortress and Gatekeeper: Theorizing Transitive Trust in Third-Party Cybersecurity Risk Governance
arXiv · 25 June 2026
How to cite this record
ethics.ai (17 July 2026), “Innocuous-Seeming Data, Latent Ideology: Ideological Generalisation in Finetuned LLMs,” evidence record 10983, https://ethics.ai/record/10983 (originally published by arXiv cs.CY).
Use and limitations
This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.