EU AI Act Compliance for Creative Industries in Italy: A Complete 2026 Guide for Film, Music, Art, and IP Professionals

The EU AI Act (Regulation 2024/1689) entered force in August 2024 with staggered application dates running through 2027. By 2026, the most consequential obligations for creative industries are already operational: synthetic content labelling, foundation model transparency, prohibited practices, and the Italian-specific protections of Italian Law 132/2025. For producers, musicians, artists, brand owners, and IP rights holders operating in Italy, AI compliance is no longer optional — it is integrated into tax credit access, distribution agreements, talent contracts, and platform terms.

This guide explains the AI compliance framework for Italian creative industries. For the broader copyright framework, see our master pillar guide. For platform liability under DSA, see our platform and intermediary liability guide.

The AI Act framework

The AI Act operates through a risk-based classification:

• Prohibited practices (Article 5): social scoring, exploitation of vulnerabilities, untargeted biometric data scraping for facial recognition databases — banned outright;
• High-risk AI systems (Annex III): substantial compliance obligations including conformity assessment, technical documentation, and ongoing monitoring;
• Limited-risk AI systems: transparency obligations including disclosure that users are interacting with AI;
• Minimal-risk AI: voluntary codes of conduct;
• Foundation models / general-purpose AI: specific obligations under Articles 51-55 including transparency, copyright respect, technical documentation, and (for systemic risk models) extensive risk management.

Most creative industries’ AI use falls under limited-risk transparency obligations and the foundation model framework. High-risk classifications are rare for creative applications.

Italian Law 132/2025

Italy was among the first EU Member States to enact national-specific AI legislation supplementing the AI Act. Law 132/2025 establishes:

• Personality rights protections: specific safeguards against unauthorised AI use of identifiable persons’ image, voice, and likeness, particularly in synthetic content contexts;
• Sector-specific provisions: integration with audiovisual, healthcare, public administration, and employment frameworks;
• Coordination with existing Italian rights: Articles 96-97 LDA (image rights), Article 10 Italian Civil Code (privacy), GDPR Article 6 (personal data processing);
• Enforcement: through Italian Data Protection Authority (Garante), AGCOM for audiovisual content, and Italian courts for civil claims;
• Penalties: substantial fines for non-compliance, with criminal sanctions for serious violations.

The Italian framework provides creative professionals with national-specific tools alongside the EU-level AI Act framework.

AI training and copyright

One of the most contested compliance areas: training AI models on copyrighted content. The framework combines:

• DSM Directive Article 3 (Italian Articles 70-ter LDA): TDM exception for scientific research, no opt-out;
• DSM Directive Article 4 (Italian Article 70-quater LDA): TDM exception for general AI training, with rights holder opt-out;
• Opt-out mechanisms: rights holders must signal opt-out in machine-readable form (robots.txt directives, metadata, platform terms);
• AI Act Article 53: foundation model providers must publish summaries of training data and respect EU copyright;
• Output liability: AI providers and users face liability where AI generates output infringing specific copyrighted works.

For Italian creators (authors, photographers, musicians, artists), implementing effective opt-out signals is essential to preserve rights against unauthorised AI training. For AI-using businesses, documentation of training data sources, opt-out respect, and output filtering provides compliance defence.

Synthetic content and labelling

AI Act Article 50 establishes the synthetic content transparency framework:

• AI-generated or manipulated audio, image, video, and text content must be clearly labelled as artificially generated;
• Deepfake content (manipulated audio/video of real persons) requires explicit disclosure;
• Labelling must be machine-readable and detectable;
• Exceptions exist for artistic, creative, satirical, or fictional use, but the underlying creator/distributor still has documentation obligations;
• Platforms hosting AI-generated content must implement detection and labelling mechanisms.

For Italian productions accessing the cinema tax credit, the mandatory AI clause under Article 7 paragraph 6 D.I. MiC-MEF 225/2024 integrates these transparency requirements into core production contracts.

Deepfake and personality rights

Deepfake content engages multiple legal frameworks simultaneously:

• AI Act Article 50: synthetic content labelling;
• Italian Law 132/2025: national-specific personality rights protections;
• Article 10 Italian Civil Code + Articles 96-97 LDA: traditional image rights framework;
• GDPR Article 6: image is personal data, requiring legal basis for processing;
• Italian defamation provisions: deepfakes presenting persons in damaging contexts engage Article 595 Italian Criminal Code;
• Platform liability under DSA: platforms hosting deepfake content face takedown obligations.

For Italian celebrities, athletes, politicians, and other public figures, comprehensive deepfake protection requires coordinated strategy across these frameworks. See our right to image guide for the personality rights framework.

Sector applications: film, music, art

Film and audiovisual: AI clause integration in director, screenwriter, composer, and performer agreements per D.I. 225/2024. AI-generated promotional materials require synthetic content labelling. See our film production IP guide.

Music: AI-generated music raises authorship and ownership questions. Voice cloning of singers and performers requires specific consent under personality rights framework. AI tools used in production must respect training data opt-outs. Distribution platforms increasingly require AI disclosure.

Art and photography: AI image generation tools face training data liability for outputs resembling protected works. Artists can implement opt-out signals on portfolio platforms. AI-generated art faces uncertain copyright protection — purely AI-generated works may lack copyright; substantial human creative contribution is required. See our AI photography guide and OpenAI Sora analysis.

Trademark and brand: AI-generated brand content (logos, packaging, advertising) faces uncertain trademark protection. AI tools used in design must respect training data opt-outs from competitor brands. AI-generated counterfeits require new enforcement approaches.

Enforcement and penalties

AI Act penalties are substantial:

• Prohibited practices violations: up to €35 million or 7% of global turnover;
• High-risk system violations: up to €15 million or 3% of global turnover;
• Other obligations: up to €7.5 million or 1.5% of global turnover;
• Italian Law 132/2025: additional national penalties for personality rights violations;
• DSA platform sanctions: up to 6% of global turnover for very serious platform violations.

For Italian creative businesses, AI compliance is not a theoretical exposure — penalties are operational risks requiring proactive management.

How DANDI supports AI compliance

DANDI.media supports Italian and international creative businesses on AI compliance:

• AI Act compliance audit and gap analysis;
• Italian Law 132/2025 implementation;
• AI clause integration in creative talent contracts;
• Training data documentation and opt-out implementation;
• Synthetic content labelling and disclosure programmes;
• Deepfake protection for talent, athletes, and public figures;
• Platform DSA compliance for AI-related content;
• Disputes and enforcement coordination.

For consultation, book directly with Avv. Claudia Roggero or Avv. Donato Di Pelino.

Related guides

Frequently asked questions

When does the EU AI Act apply to my creative business?

The AI Act applies progressively from February 2025 (prohibited practices) through August 2027 (full application). By 2026, synthetic content labelling, foundation model obligations, and Italian Law 132/2025 are all operational.

Do I need to label AI-generated marketing content?

Yes, for substantial AI-generated or AI-manipulated content. Article 50 AI Act requires labelling of synthetic audio, image, video, and text. Limited exceptions for artistic and satirical use exist but require careful assessment.

Can I use AI to train on my competitors’ content?

Not freely. The DSM Article 4 TDM exception allows training but only where rights holders have not opted out. Italian Articles 70-ter and 70-quater LDA implement this with specific opt-out requirements. Output infringement liability also applies.

What protects me against deepfake use of my image?

Italian Law 132/2025 plus Articles 96-97 LDA, Article 10 Italian Civil Code, GDPR Article 6, and AI Act Article 50 collectively provide comprehensive protection. Enforcement through Garante, AGCOM, and Italian courts.

Dandi Law Firm provides legal assistance in several Practice Areas. Check out our Services or contact Us!