Defending Against Neural Fake News

TL;DR

This paper introduces Grover, a controllable text generation model for neural fake news, evaluates detection methods, and discusses ethical implications, emphasizing the importance of releasing strong generators for improved detection.

Contribution

The paper presents Grover, a new model for generating neural fake news, and analyzes detection techniques, highlighting the effectiveness of using the generator itself as a defense.

Findings

Discriminators achieve 73% accuracy in detecting neural fake news.

Grover can generate highly trustworthy-looking fake news articles.

Using Grover as a detector yields 92% accuracy, emphasizing the importance of releasing strong generators.

Abstract

Recent progress in natural language generation has raised dual-use concerns. While applications like summarization and translation are positive, the underlying technology also might enable adversaries to generate neural fake news: targeted propaganda that closely mimics the style of real news. Modern computer security relies on careful threat modeling: identifying potential threats and vulnerabilities from an adversary's point of view, and exploring potential mitigations to these threats. Likewise, developing robust defenses against neural fake news requires us first to carefully investigate and characterize the risks of these models. We thus present a model for controllable text generation called Grover. Given a headline like `Link Found Between Vaccines and Autism,' Grover can generate the rest of the article; humans find these generations to be more trustworthy than human-written disinformation. Developing robust verification techniques against generators like Grover is critical. We find that best current discriminators can classify neural fake news from real, human-written, news with 73% accuracy, assuming access to a moderate level of training data. Counterintuitively, the best defense against Grover turns out to be Grover itself, with 92% accuracy, demonstrating the importance of public release of strong generators. We investigate these results further, showing that exposure bias -- and sampling strategies that alleviate its effects -- both leave artifacts that similar discriminators can pick up on. We conclude by discussing ethical issues regarding the technology, and plan to release Grover publicly, helping pave the way for better detection of neural fake news.