Viable Threat on News Reading: Generating Biased News Using Natural Language Models

TL;DR

This paper demonstrates that publicly available natural language models can reliably generate biased news articles, which are fluent and easily identifiable, posing a threat to news integrity and reader perception.

Contribution

It introduces a threat model showing how controllable language models can produce high-quality biased news, highlighting security concerns in natural language generation.

Findings

Generated biased news is fluent and coherent.

Bias in generated articles is clearly identifiable.

High-quality biased news can be systematically produced.

Abstract

Recent advancements in natural language generation has raised serious concerns. High-performance language models are widely used for language generation tasks because they are able to produce fluent and meaningful sentences. These models are already being used to create fake news. They can also be exploited to generate biased news, which can then be used to attack news aggregators to change their reader's behavior and influence their bias. In this paper, we use a threat model to demonstrate that the publicly available language models can reliably generate biased news content based on an input original news. We also show that a large number of high-quality biased news articles can be generated using controllable text generation. A subjective evaluation with 80 participants demonstrated that the generated biased news is generally fluent, and a bias evaluation with 24 participants demonstrated that the bias (left or right) is usually evident in the generated articles and can be easily identified.