Italy’s antitrust authority closes probes into DeepSeek, Mistral, and Nova AI over AI hallucination disclosures

Italy’s antitrust authority, the AGCM, has concluded its investigations into prominent AI chatbot providers DeepSeek, Mistral AI, and Nova AI. This development is significant as it sets a precedent for how AI models must disclose their limitations, particularly concerning "hallucinations" or fabricated information. The AGCM’s decision establishes a tangible standard for transparency in the AI sector, impacting how generative AI services will be presented to consumers across Europe. The AGCM secured binding commitments from all three companies, mandating improved disclosure of AI hallucination risks. These commitments provide a clear benchmark for what constitutes adequate transparency in practice, with a 120-day compliance period before potential penalties are enforced. This move by Italian regulators highlights a growing European focus on ensuring factual accuracy and user understanding of AI-generated content. This regulatory action directly affects users of these AI services, as well as the broader AI development community, by imposing stricter requirements for disclosure. The AGCM’s stance underscores a commitment to consumer protection within the rapidly evolving digital landscape, particularly concerning complex technologies like artificial intelligence. It signals a potential shift towards greater accountability for AI providers in addressing the inherent challenges of current AI capabilities.