Russia-linked network wages AI disinformation war across globe

A Russia-affiliated disinformation network called Portal Kombat has been identified spreading coordinated false narratives across multiple platforms to poison the data sources that train major AI models including ChatGPT, Grok and Gemini, according to intelligence officials. The operation exploits a fundamental vulnerability in how large language models learn—by absorbing information from vast swaths of internet content—to inject propaganda at scale. By seeding misleading content across hundreds of websites, the network aims to subtly shift how AI systems respond to queries about geopolitical events, military conflicts and other strategically sensitive topics.

The discovery underscores a growing challenge as artificial intelligence systems become more influential in shaping public discourse and decision-making. Unlike traditional disinformation campaigns targeting human audiences directly, this approach weaponizes the machine-learning process itself, making detection and mitigation significantly more complex. The technique requires no viral spread or mass audience engagement to be effective—it simply needs to be present in the training data that AI developers harvest from the web.

Experts emphasize that combating AI-focused disinformation will require coordinated action across technology companies, intelligence agencies and content platforms. Solutions being discussed include improving transparency in AI training data sources, developing detection systems for coordinated inauthentic behavior at scale, and establishing international norms against using AI systems as vectors for state-sponsored information warfare. The Portal Kombat case demonstrates that adversaries are rapidly adapting their tactics to exploit emerging technologies, even as AI itself becomes a potential tool for identifying and countering such operations.