L'IA peut-elle mener des campagnes de guerre psychologique illimitées sur les réseaux sociaux à grande échelle ?

Les systèmes d'IA génèrent et déploient déjà de la désinformation ciblée avec une efficacité alarmante. La technologie existe pour créer de la propagande personnalisée pour des millions d'individus simultanément. Les plateformes existantes peinent à détecter et contrer ces campagnes pilotées par l'IA en temps réel. L'étape suivante serait des opérations psychologiques entièrement autonomes sans aucune intervention humaine.

Background

AI systems already generate and deploy targeted disinformation with alarming effectiveness. The technology exists to create personalized propaganda for millions of individuals simultaneously. Existing platforms struggle to detect and counter these AI-driven campaigns in real-time. The next step would be fully autonomous psychological operations with no human intervention. While AI can analyze large datasets to identify human vulnerabilities, generate tailored content, or automate social media interactions, its deployment in warfare is heavily restricted by ethical guidelines, legal frameworks, and platform policies. Existing tools are primarily used for defensive applications (e.g., detecting disinformation) rather than offensive operations. Governments and militaries explore such technologies under classified programs, but widespread, unchecked use remains prohibited by international conventions like the Tallinn Manual and the Geneva Convention protocols. Models like those in the Meta AI and OpenAI portfolios can generate human-like content and interact with users in a way that can influence opinions and emotions. The effectiveness and ethics of such campaigns are highly debated, and their use is often considered controversial. The current state of the art in natural language processing and machine learning has made it possible for AI to automate and amplify psychological warfare tactics on social media.