Italy’s antitrust regulator is tightening consumer rights and transparency in AI services, policing unfair practices related to generative AI.

The AGCM’s action underscores its role in safeguarding consumers from misleading AI claims and in promoting responsible AI deployment with clear user awareness of AI limitations.

Firms must display permanent in-chat disclaimers about hallucination risks on websites and apps to clearly inform users.

In addition, companies are required to communicate the risks of AI hallucinations where users interact with the technology.

The companies involved are DeepSeek (China), Mistral AI SAS (France), and Scaleup Yazilim Hizmetleri Anonim Şirketi (Turkey).

Devdiscourse reported the developments on April 30, 2026, with input from authorities.

Investigations focused on potentially unfair practices tied to generative AI and the risk of hallucinations, which produce inaccurate or misleading content.

Italy’s AGCM closed the probes into three AI firms after they accepted binding commitments to address hallucination risks and improve real-time disclosures.

The regulator frames in-use warnings as a core consumer protection obligation to prevent harm from overreliance in high-stakes domains.

The closure emphasized transparency and consumer information as central to resolving the probes without penalties.

DeepSeek pledged to invest in technology to reduce hallucinations, while acknowledging current limits in preventing them.

Scaleup’s NOVA AI service will disclose its role as an aggregator of multiple models and explicitly state it does not process or aggregate user responses.