AI Data Poisoning Distorts WWII History in Large Language Models

Thumbnail Image

The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.

Chinese officials accused Japanese right-wing groups of deliberately poisoning training data for international AI language models with falsified WWII historical narratives. This manipulation has caused AI systems to output distorted accounts, harming communities by spreading misinformation and undermining historical truth. China urges global vigilance and resistance against such misuse of AI.[AI generated]

Why's our monitor labelling this an incident or hazard?

The event explicitly involves AI systems (large language models) whose training data has been deliberately poisoned to produce outputs that distort historical facts. This use of AI leads to the dissemination of false narratives about significant historical events, causing harm to communities by undermining historical truth and potentially fueling harmful political agendas. The AI system's outputs are directly linked to the harm, fulfilling the criteria for an AI Incident involving violations of rights and harm to communities.[AI generated]
AI principles
Robustness & digital securityTransparency & explainability

Industries
Digital securityMedia, social platforms, and marketing

Affected stakeholders
General publicCivil society

Harm types
Public interest

AI system task:
Content generation


Articles about this incident or hazard