Italy

[IT] Italy enacts comprehensive AI law that establishes human authorship requirement for copyright protection and criminalises deepfake dissemination

IRIS 2025-10:1/10

Ernesto Apa and Chiara Marchisotti

Portolano Cavallo

On 23 September 2025, Italy enacted a new law designed to facilitate the application of the AI Act. The law promotes correct, transparent and responsible AI use whilst guaranteeing oversight of risks and impact on fundamental rights. Within the broader framework of this law, a few copyright-related provisions merit particular attention. 

The most significant intellectual property provision appears in Article 25, which amends the Italian Copyright Law to insert the word “human” into the phrase “works of intellect”, and to specify that works created with the assistance of artificial intelligence tools qualify for protection only when “they constitute the result of the author’s intellectual work”. This establishes an explicit human element requirement for copyright protection, a standard that will evidently develop through case law, as courts distinguish between varying levels of human involvement in the creative process.

Simultaneously, the law also introduces Article 70-septies to the Italian Copyright Law permitting reproductions and extractions from legitimately accessible works for text and data mining through AI systems, including generative AI, whilst preserving protections under the Berne Convention.

The Italian AI Law also introduces criminal sanctions targeting AI-enabled misconduct, with the centrepiece being a new provision of the Italian Criminal Code establishing that anyone who causes unjust harm to a person by disseminating, without their consent, images, videos or voices falsified or altered through the use of artificial intelligence systems and non-consensual deepfakes capable of misleading as to their genuineness, faces imprisonment of between one and five years. Prosecution generally requires a victim complaint, except when connected to offences or involving vulnerable persons (such as minors) or public officials. 

The new law establishes AI-specific aggravating circumstances and enhanced penalties across multiple criminal domains, including a general aggravating circumstance when AI systems constitute insidious means or aggravate the consequences of a crime. Political conspiracy involving AI-based deception is punishable by two to six years' imprisonment, whilst AI-based market manipulation attracts two to seven years plus fines of up to EUR 6 million. 

Unauthorised text and data mining from online works using AI systems also constitutes criminal copyright infringement, whilst preserving legitimate access exceptions. 

The law presents both opportunities and challenges for different stakeholders. Rights holders gain clearer legal frameworks to protect human creative works and prevent unauthorised AI training, whilst developers and platforms obtain greater certainty regarding permissible uses under the legitimate access framework. Both constituencies face implementation questions as courts develop interpretative guidance on key provisions.


References





This article has been published in IRIS Legal Observations of the European Audiovisual Observatory.