AI | AGCM closes investigation into DeepSeek following commitments on AI hallucinations

On 5 January 2026, the Italian Competition Authority (AGCM) closed the proceedings against Hangzhou DeepSeek Artificial Intelligence Co., Ltd. and Beijing DeepSeek Artificial Intelligence Co., Ltd. (together, DeepSeek), without finding any infringement, after DeepSeek agreed to a set of binding commitments aimed at enhancing transparency and consumer awareness. The investigation concerned the alleged insufficient disclosure to users of the risk that DeepSeek’s AI models may generate inaccurate or misleading outputs (“hallucinations”), which could potentially constitute an unfair commercial practice under the Italian Consumer Code. In particular, DeepSeek undertook to introduce permanent informational banners in Italian within the chat interface and on the sign-up page, warning users that the content is AI-generated and may be inaccurate, with direct links to Italian-language terms of use and the privacy policy. Additional warnings will be displayed in responses generated in sensitive areas (including medical, legal and financial topics). DeepSeek also committed to providing a full Italian translation of its terms of use, as well as improving its AI models through data filtering, post-training reinforcement techniques and the use of real-time authoritative sources.

 Newsletter n. 114 – January 2026