AI Decree: new criminal and corporate liability rules for businesses from 30 September

by Francesco Rubino

Legislative Decree No. 160 of 9 September 2026 comes into force on 30 September 2026, bringing national legislation into line with Regulation (EU) 2024/1689 (“AI Act”) concerning the use of artificial intelligence systems in policing and civil and criminal liability.

The new criminal offence

The legislation punishes those who fail to adopt the technical safety measures required for the design, training, production and placing on the market of high-risk AI systems – measures suitable for preventing malfunctions or tampering – or measures involving human oversight, in cases where this poses a danger to life or to public or individual safety, or to national security.

The provision also punishes anyone who tampers with high-risk artificial intelligence systems, where this results in a danger to life, public or individual safety, or national security.

The provision also provides for criminal liability for omissions committed through gross negligence.

A professional who intentionally fails to put in place measures for human supervision shall also be held liable.

Legal liability of organizations is now also provided for in relation to "offences committed through the use of artificial intelligence systems", and this newly introduced provision is included amongst the offences giving rise to the liability of organizations.

Organizations are also liable for offences relating to the unlawful dissemination of images, videos or voices that have been falsified or altered using artificial intelligence (deep fakes).

Practical implications for Organizational Models

The new provisions are of general application and apply to any business, organisation or professional involved in the design, training, production, placing on the market or professional use of high-risk AI systems.

Businesses and professionals are therefore required to assess whether it is necessary to update their Organizational Model and the procedures for human oversight of high-risk AI systems.