Italy’s competition authority, the Italian Competition Authority, has closed investigations into several artificial intelligence companies after they agreed to new transparency measures aimed at reducing risks linked to AI “hallucinations” — inaccurate or misleading outputs generated by language models. The probes involved Mistral AI, DeepSeek, and Scaleup Yazilim Hizmetleri. The regulator said the firms accepted binding commitments to improve how they inform users about the limitations of generative AI systems. These include adding permanent disclaimers in chatbot interfaces and clearer explanations on websites and applications about the possibility of incorrect outputs. Some companies also pledged technical improvements, with DeepSeek agreeing to invest in methods to reduce hallucination risks, though acknowledging that such errors cannot be fully eliminated with current technology. Scaleup’s chatbot service will also clarify that it acts as an interface to multiple AI systems rather than generating unified responses.
Read more here
