Why Is AI Not Perfect? Regulators Are Forcing Chatbots to Admit the Flaw
Italy’s competition authority has required DeepSeek, Mistral, and NOVA AI to warn users about AI “hallucination” risks, treating it as a consumer-protection issue. The move follows rapid adoption, with generative AI reaching 53% of the population and 88% of organizations in Italy, according to Stanford HAI. EU obligations for general-purpose AI models are already in force, with enforcement powers coming August 2, 2026.