On 23 April, the Italian Council of Ministers/Cabinet approved a bill for the introduction of measures and the delegation of powers to the Government in the field of artificial intelligence (AI).

It is therefore not yet a state law, but a government proposal that will be brought to Parliament to be discussed and possibly amended, before it is approved and enters into force.

In the Italian Government’s intentions, the scope of this bill should not overlap with that of the so-called “AI Act”, which was approved by the European Parliament on 13 March and which we have reported on in our previous newsletters, delving into the complexity of the relationship between AI and copyright and the limits of the use of biometric recognition technologies.

The stated aim of this bill is mainly to regulate the use of AI-based technologies in such a way as to minimise the risks for citizens, based on an “anthropocentric vision”.

Alongside a number of predictions that are expected to fuel the sector and public debate, such as the investment of “up to” a billion euro in total to support companies in the sector, the need for parental consent for the use of AI technologies for children under the age of 14, and the introduction into the legal system of the crime of unlawful dissemination of content generated or manipulated with AI systems, there are also proposals that directly affect the Intellectual Property sector.

In detail, the government draft proposes amendments to the Copyright Act and to the law regulating audiovisual media.

The bill envisages amending Art. 1 of the Copyright Act by limiting protection to works of human genius only, specifying “even where created with the aid of artificial intelligence tools, provided that the human contribution is creative, relevant and demonstrable”.

In practice, “copyright” cannot be claimed for a text (including the listing of a computer program), a film or an image made only by a machine or with negligible human contribution, even if the same work would have been eligible for protection if it had been created by a person. On this point, Italy would thus go in a direction consistent with that of other Western countries, such as the United States, for example.

If the amendment is approved, for the purposes of acquiring authorial protection for AI creations, it will therefore become crucial to prove that there was also a creative, relevant and demonstrable human contribution, which will predictably fuel the debate among specialised legal experts and, sooner or later, also in the Courts.

In addition, it was decided to introduce an article to the Copyright Act that would extend the right of citation and free use of protected works to include the “reproduction and extraction of works or other materials by means of models and systems of (generative) artificial intelligence”.

Today, the law already permits the summary, quotation or free partial reproduction of a protected work, as well as its publication, if carried out for the purpose of research, criticism or discussion, provided that such use is not in commercial competition with the work itself. Extending this exception to author protection also to generative models, such as Chat-GPT, Midjourney, etc., which are now part of our everyday life, could make it complicated to sanction abuses.

Also of great interest is the proposal to amend the Consolidated Text on Audiovisual Media Services, so as to extend the ban on the use of methodologies and techniques that are capable of manipulating content in a way that is unrecognisable to the viewer  to the use of AI systems, and introduce an obligation to mark with the abbreviation “AI” any video content that has been, even partially, generated by (or intervened upon by) an artificial intelligence system. In the case of audio-only content, there would be an obligation to report “by means of audio announcements or technologies suitable for recognition”.

One of the purposes of this change would seem to be to curb the negative effects of so-called “deep fakes”. However, any content that is “manifestly creative, satirical, artistic or fictitious” would be excluded from the obligation to display the “AI label”.