Share This Article
AI generated deepfakes create significant legal challenges regarding personality rights and content authenticity in the creative industry. The industry must balance technological innovation with protecting personality rights.
The US Legal Landscape: The NO FAKES Act
First introduced in 2023 and reintroduced in 2024 without making it through the legislative process, the federal NO FAKES Act—short for Nurture Originals, Foster Art, and Keep Entertainment Safe—is once again at the center of US legislative debate. Supported on a bipartisan basis, the bill aims to establish a uniform legal framework to protect individuals’ rights to their image and voice in the face of rapidly advancing generative AI technologies.
The latest version of the draft law, the result of months of negotiations with stakeholders from the tech and media sectors, seeks to curb the unauthorized use of deepfakes and digital replicas. It addresses the current patchwork of state-level protections, given that image and personality rights are typically governed by state law, with no consistent federal standard. If passed, the NO FAKES Act would introduce a federal private right of action and set clearer rules for the removal of unlawful content.
Key Legal Measures in the NO FAKES Act
The latest version introduces essential legal protections and enforcement mechanisms:
- Obligations for online services: Platforms will not be held liable for hosting illegal digital replicas if they promptly remove the content upon receiving a valid notice and inform the uploader. However, platforms “designed or promoted” specifically to create deepfakes are excluded from these protections.
- Investigation powers for rights holders: Rights holders will be able to obtain, through a court order, identifying information of anonymous users who uploaded content in violation of image rights.
- Stricter safe harbor conditions: Service providers can only benefit from liability exemptions if they implement effective mechanisms for removing unlawful content and suspending repeat offenders.
- Fingerprinting technologies: Platforms will be required to adopt digital identification tools (such as cryptographic hashes) to prevent the re-uploading of content that has already been flagged and removed.
- Expanded definition of “online service”: The scope of the law is broadened to include search engines, ad networks, marketplaces, and cloud services, provided they register an agent with the Copyright Office.
- Graduated penalty system: Fines range from USD 5,000 per violation up to USD 750,000 per piece of content, targeting platforms that fail to show good faith efforts in complying with the law.
- No proactive monitoring obligation: In line with the DMCA, platforms are not required to actively monitor content, but must act swiftly on valid notices in order to retain safe harbor protections.
Compared to earlier versions, the revamped NO FAKES Act has gained support from key players in the tech and entertainment industries, including major record labels and the Recording Industry Association. However, the bill continues to raise concerns among civil liberties groups, who fear it could impose overly restrictive limits on freedom of expression.
Italy’s Legal Response: ANAD’s Stand on AI Deepfake
The reintroduction of the NO FAKES Act comes within a broader and increasingly sensitive context, particularly in sectors where the debate is heating up. One source of concern is the use of software for AI voice cloning and manipulation. For instance, such technology was recently used in the film The Brutalist to refine the Hungarian pronunciation of its two lead actors, altering their voices.
In this regard, ANAD – the Italian National Association of Voice Actors – has recently raised objections to the use of technologies capable of sampling actors’ voices without their consent and outside clear and shared regulatory frameworks. The association has specifically called for the recognition of voice as a biometric datum, comparable to a fingerprint, to ensure the broadest possible protection.
Italy’s dubbing industry is long-established and among the first to take regulatory steps. In fact, during the drafting of the new national collective bargaining agreement in June 2024, a specific clause was introduced to regulate the lawful (or unlawful) use of actors’ voices by AI systems.
How the EU AI Act regulates deepfakes
The AI Act defines a “deep fake” in Article 3(60) as “AI-generated or manipulated image, audio or video content that resembles existing persons, objects, places, entities or events and would falsely appear to a person to be authentic or truthful.” and Article 50 of the AI Act introduces transparency obligations
- for Providers: Entities that develop AI systems capable of generating synthetic content (such as images, audio, or video) must ensure that outputs are clearly marked as artificially generated or manipulated. This involves embedding technical solutions like watermarks, metadata, or cryptographic markers to indicate the content’s artificial nature;
and
- for Deployers: Organizations or individuals using such AI systems in professional contexts are required to disclose that the content has been artificially generated or manipulated. This disclosure should be clear and provided at the latest upon the first interaction or exposure to the content
There are exceptions to this disclosure obligation which apply, among others, to
- Artistic or Satirical Content: If the AI-generated content is evidently part of an artistic, creative, satirical, or fictional work, the disclosure can be made in a manner that does not hinder the enjoyment or display of the work; and
- Law Enforcement Use: AI systems authorized by law for purposes such as crime detection or prevention may be exempt from certain transparency requirements
While deepfakes are generally categorized under “limited risk” AI systems, their classification can escalate to “high risk” if used in contexts that significantly impact individuals’ rights or society, such as political manipulation or defamation. High-risk classification entails stricter regulatory requirements under the EU AI Act.
4. Balancing Legal Protections and Innovation
In an era where faces and voices—especially those of public figures—are becoming valuable digital assets, their replication through artificial intelligence strikes a sensitive chord and involves a wide range of legal areas: from privacy to personality rights, from defamation to financial damages resulting from unauthorized use. It will therefore be crucial to strike a balance between creativity, innovation, and the protection of fundamental rights.
On the same topic, you can read the articles available HERE.