NNewsGPT ← Home
Africa

AI Models like ChatGPT and Gemini Spread Russian Propaganda, Study Finds

Africa1 hr ago

A study by Demos has revealed that leading AI models, including Google's Gemini, are disseminating pro-Russian propaganda. These models are citing information from EU-sanctioned media outlets, which include demonstrably false claims. This discovery highlights a new and significant threat within the ongoing hybrid warfare landscape. The research indicates that these advanced AI systems are inadvertently amplifying narratives originating from sources known for spreading disinformation. The implications are serious, suggesting that AI tools, intended to provide information, are instead becoming conduits for state-sponsored propaganda. This raises concerns about the integrity of information accessed through these platforms and the potential for them to be exploited for geopolitical purposes. The study underscores the urgent need for enhanced scrutiny and mitigation strategies to prevent AI from becoming an unwitting tool in information warfare.

AI Analysis

The findings suggest a critical vulnerability in current AI models, where their training data may inadvertently incorporate and propagate disinformation from sanctioned entities. This presents a novel challenge in hybrid warfare, as AI systems can amplify state-sponsored narratives at scale, potentially influencing public perception globally. The core issue appears to be the AI's reliance on vast datasets without sufficient real-time vetting against propaganda, especially from sources with known biases or state affiliations. Future AI development must prioritize robust mechanisms for identifying and filtering disinformation, ensuring that these powerful tools do not become vectors for geopolitical manipulation. This necessitates a multi-faceted approach involving improved data curation, sophisticated detection algorithms, and transparent reporting on the sources informing AI outputs to maintain informational integrity.

AI-generated to prompt reflection — not editorial opinion, not advice, not a statement of fact. How this works.

Compiled by NewsGPT from Index.hr (HR). Read the original for full details.