Italy's Antitrust Authority Closes Probes into DeepSeek, Mistral, and Nova AI over AI Hallucination Disclosures
April 30, 2026 - 10:55 am
The AGCM (Italy’s competition and consumer protection authority) has closed its investigations into three AI chatbot providers—China’s DeepSeek, France’s Mistral AI, and Turkey’s Scaleup Yazilim (operator of Nova AI)—after each company agreed to binding commitments aimed at improving user warnings about AI hallucinations.
Background
The AGCM opened probes into the companies based on their failure to clearly, immediately, and intelligibly inform users that their AI chatbots could generate inaccurate, misleading, or fabricated content. This lack of transparency was deemed a potentially unfair commercial practice under Italy’s Consumer Code, hindering users from making informed decisions, especially in high-stakes areas like health, finance, and law.
Resolutions
None of the cases resulted in formal findings or fines. Instead, all three were resolved through the commitment mechanism available under Article 27(7) of the Consumer Code, where companies propose remedies to address the authority’s concerns. The AGCM accepted these proposals, which included:
- DeepSeek: Prominent hallucinations risk warnings on chat interfaces and websites, full Italian translations of relevant disclosures, internal compliance training workshops, and a commitment to invest in reducing hallucination rates.
- Mistral AI: In-chat disclaimers ("Le Chat may make mistakes. Please check responses"), strengthening and translating existing warnings, and other measures.
- Nova AI: Similar commitments as Mistral AI, tailored to their specific transparency failures.
Non-compliance with these commitments within a 120-day window will reopen the cases and expose each company to fines of up to approximately $11.6 million.