Existing clients

Log in to your client extranet for free matter information, know-how and documents.

Client login

Staff

Mills & Reeve system for employees.

Staff Login
19 Nov 2026
6 minutes read

Deepfakes, reputations and the limits of the law

A fake endorsement. A cloned voice. A video showing someone saying something they never said. What do existing laws have to say? And is it enough?

As artificial intelligence becomes more sophisticated, the creation of convincing "digital replicas" of real people is becoming easier than ever. What was once the preserve of specialist studios can now be produced using widely available tools and shared online within minutes.

For professional athletes, broadcasters and entertainers, the risks are obvious. A fabricated interview or endorsement can undermine carefully managed commercial relationships and damage years of brand building. Recent examples demonstrate the scale of the problem. In September 2026, an AI-generated fake interview appearing to show Kylian Mbappé attracted more than 25 million views, misleading viewers and even some media outlets. But the issue extends far beyond public figures. Anyone with an online presence can find their image, voice or identity used without their consent.

As described in AI, deepfakes and your image rights | Mills & Reeve, current law offers some protection. The problem is that those protections were developed long before AI-generated content became a reality. This article focuses on how defamation, privacy and data protection laws fit into the current patchwork of legal remedies against unauthorised deepfakes.

Defamation: proving damage to reputation

Where a deepfake falsely depicts an individual engaging in misconduct or making offensive remarks, defamation may be one of the first causes of action considered.

To succeed, an individual would generally need to show that the publication conveys a defamatory meaning, refers to them, has been communicated to others and has caused, or is likely to cause, serious harm to their reputation. Publishers, journalists and others who repeat or share AI-generated content may also face potential exposure, not just the original creator.

In practice, however, defamation can be an imperfect solution to the deepfake problem. Content may spread rapidly across multiple platforms, often anonymously and sometimes internationally. By the time legal action is contemplated, the damage may already have occurred and the original source may be difficult to identify. There are also various defences to a defamation claim, which might come into play.

Defamation law also focuses on harm to reputation. It does not directly address the separate concern that someone has misappropriated an individual's face, voice or identity in the first place.

Privacy: protecting the individual, not necessarily the likeness

Privacy law may assist in some circumstances. The courts can protect information where there is a reasonable expectation of privacy and where the individual's privacy rights outweigh competing rights to freedom of expression.

The difficulty is that many deepfakes do not necessarily disclose private information. A fabricated video may be entirely fictional. The harm lies not in the disclosure of confidential information, but in the creation of a convincing imitation.

This creates a potential gap. Privacy law was developed to prevent inappropriate disclosure of private material. It was not designed as a comprehensive mechanism for controlling the use of a person's identity.

Data protection: an increasingly important battleground

Data protection law may be one of the most relevant legal regimes in the AI context.

Creating a convincing deepfake generally requires photographs, video recordings, voice samples or other information relating to an identifiable individual. That information is likely to constitute personal data. Questions also arise as to whether some of the information used to train or develop these systems could constitute biometric data, requiring additional protections and making compliance more complex.

But an individual does not “own” their personal data. Instead, data protection law gives individuals rights over how their personal data is processed and imposes corresponding obligations on those using it. The key question is therefore generally whether the personal data used to create a digital replica has been processed lawfully, rather than who “owns” that data.

The fact that photographs, recordings or other information are already publicly available online does not automatically make their reuse lawful. AI developers and content creators must identify a lawful basis for processing that data and consider the impact on affected individuals. Creating a digital replica without the individual’s consent is not necessarily unlawful under data protection law. Consent is only one potential lawful basis for processing personal data. Whether processing is lawful will depend on the circumstances, including the data being used, the purpose and lawful basis relied upon, what the individual has been told about the processing, and whether that processing is fair and proportionate.

Where a developer seeks to rely on "legitimate interests" (as a lawful basis for processing), it must balance its own objectives against the rights and freedoms of the individual, including the risk of reputational harm, distress and loss of control over their personal data.

Therefore, the lawfulness of a deepfake under data protection law can come down to a case-by-case analysis of how it was created, using what data, on what basis, in what context and for what purposes. This can be very difficult to ascertain as the processing behind AI-generated content is often invisible to the individuals concerned, who may have no idea that their images, recordings or online content are being used to create digital replicas.

So data protection law is not a straightforward answer either. While individuals may be able to challenge unlawful processing, complain to the ICO or seek compensation where damage or distress has been suffered, the legislation was not designed to regulate digital replicas. More fundamentally, data protection law regulates the processing of personal data. It does not create a standalone right allowing individuals to control every use of their image, voice or likeness.

A patchwork of protections

Taken together, defamation, privacy and data protection law can provide valuable remedies in the right circumstances. The challenge is that each addresses a different aspect of the problem. Defamation protects reputation. Privacy protects private information. Data protection regulates the use of personal data - having rights over the processing of personal data is not the same as having a proprietary right in your identity or in every digital replica created from it. Even a successful claim may address only a particular use, publication or defendant, rather than preventing the same person's likeness being replicated again elsewhere. None was specifically designed to answer a simple question: Who should control an individual's name, image, voice and likeness in the age of artificial intelligence?

Recent developments suggest that regulators and policymakers are becoming increasingly alive to the risks posed by AI-generated content. The ICO has repeatedly emphasised that existing data protection laws apply to AI systems and the personal data used to create digital replicas, while Ofcom's powers under the Online Safety Act are increasingly being used to address harmful deepfakes and other forms of AI-generated abuse online. The Government has also acknowledged concerns that existing laws may not give individuals meaningful control over their image and voice. Yet these developments also highlight the limits of the current framework: while various laws may provide remedies in particular circumstances, there remains no clear, standalone right allowing individuals to control the use of their name, image, voice and likeness.

That gap sits at the heart of the My Name, My Face campaign My Name, My Face - Campaign for UK personality rights legislation. The campaign calls for dedicated personality rights legislation, giving individuals greater control over how their identity is used. While public figures are often the most visible examples, the issues raised by AI-generated content affect everyone.

As technology continues to blur the distinction between real and synthetic content, the question is becoming increasingly difficult to ignore: should individuals have a clearer legal right to control who they are?

Our content explained

Every piece of content we create is correct on the date it’s published but please don’t rely on it as legal advice. If you’d like to speak to us about your own legal requirements, please contact one of our expert lawyers.