The reported US$975 million Khaby Lame transaction marks a genuinely new development in intellectual property (IP) transactions. The asset being commercialised does not fit neatly within the traditionally recognised categories of IP under UK or EU law. This is not the licensing or sale of pre-existing content, catalogues or brands in the conventional sense, but instead the deal involves the commercialisation of an AI-enabled version of a human identity – something more akin to digital personality rights, though with many other IP and related rights in the mix.
The transaction offers the rights to generate new AI-created performances at scale, using Lame’s vocal and behavioural data to create a continuously operating digital twin of Lame. The commercial value here lies not in the reproduction of protected works, but in the generative output.
From a UK/EU transactional IP perspective, this deal breaks new ground by authorising the use of personality rights in a generative context, enabling the creation of new, real-time and adaptive outputs. This is a context largely unfamiliar in IP transactions which are usually structured around fixed or pre-approved content. It sits at the intersection of image rights (as protected through passing off and privacy), copyright (including performers’ rights), trade marks, data protection, advertising codes, and AI-specific licensing. It raises difficult questions around ownership, control, duration and risk allocation – many of which are not expressly resolved in statute or yet in case law, and are therefore best addressed through contractual means.
In this post we look at the background to the deal, how it differs in scope to what has gone before, the core IP principles at play and their practical implications, including identity as a generative asset, control, scalability, reputational risk and AI-specific IP schedules and duration issues.
Background
Khaby Lame rose to global prominence on TikTok through silent reaction videos built around facial expression, timing and physical gesture rather than spoken language. This matters both commercially and legally. Lame’s value is not anchored in authored scripts or literary works protected by copyright, but in recognisable behavioural patterns and personal attributes, which do not attract standalone proprietary protection.
As such, what appears to be at play here is Lame authorising a broad set of rights in order to permit the training and deployment of AI models capable of replicating his likeness and mannerisms to generate new, AI-created content, particularly for livestream e-commerce.
In practical terms, this would likely involve an AI digital twin of Lame appearing on screen, presenting and reacting to products in his recognisable style, operating continuously across time zones and potentially responding dynamically to viewer prompts.
How this transaction differs
Voice and likeness licensing is already a feature of UK and EU media, advertising, sports and performance contracts. These arrangements typically rely on passing off and endorsement principles, copyright in specific recordings or performances, and/or narrowly framed contractual consents.
What differentiates this transaction is not the use of AI per se, but the scope and autonomy of the permitted exploitation. Unlike traditional endorsement or likeness deals, the output here is generative and unscripted rather than fixed and pre-approved, and it appears that it will be capable of real-time commercial interaction. The asset being licensed, and the core source of value, is not simply likeness or voice, but the ability for ‘Lame’ to perform continuously in a commercial environment.
As such, familiar drafting approaches centred on the use of specified materials or approval of finished content are insufficient. The contract must instead define and constrain permitted generative behaviour, and tightly control risk and liability.
Core principles and practical implications
(a) Identity as a generative asset
UK (and EU) law does not recognise a standalone, proprietary ‘identity right’. What is being sold and commercialised here seems to be a form of generative identity licence permitting the creation, training and commercial deployment of AI systems using Lame’s personal and behavioural attributes.
This will need to be assembled through contract from a combination of legal interests, including:
- Lame’s image and likeness (protected indirectly through passing off and misuse of private information);
- his voice and performance (to the extent protected by performers’ rights);
- personal data;
- trade marks and goodwill; and
- express contractual consent.
The contract must therefore do significant work to create exclusivity, control and enforceability where the law does not otherwise provide it.
Lame’s AI digital twin is not concerned with the reuse of existing protected works, but with the use of a system capable of generating new output derived from his attributes. The key legal question is not whether a particular output infringes copyright, but whether the scope of the AI twin’s generative behaviour has been adequately defined and constrained.
For transactional IP lawyers, the focus therefore shifts away from identifying a single controlling right, and towards drafting a contractual framework that deals with the various legal interests at play and allocates consent, exclusivity, control and liability for generative use and permitted (and prohibited) outputs.
(b) Control, scalability and reputational risk
The commercial value of Lame’s digital twin lies in scalability. The legal and personal risk for Lame lies in loss of control. This tension is particularly acute in jurisdictions such as the UK, where reputation is protected through a patchwork of indirect causes of action rather than a unified personality right.
Risk allocation is therefore critical. If Lame’s AI digital twin promotes inappropriate products, makes offensive statements or operates in sensitive markets, liability may arise through passing off, regulatory enforcement or contractual breach. In practice, this is likely to drive enhanced indemnities and negotiated liability caps; detailed brand, conduct and content covenants; termination rights linked to reputational harm; and tailored insurance considerations.
(c) AI-specific IP schedules and duration issues
From a UK/EU transactional perspective, deals of this nature highlight how generative AI is pushing at the boundaries of existing legal concepts. This is driving the use of highly tailored AI provisions to address issues such as:
- training data sources and associated legal rights;
- ownership and control of the model itself and its outputs;
- retraining and fine-tuning permissions and restrictions; and
- model deletion, retirement and post-termination use.
Digital twin arrangements also force difficult questions around duration. AI models do not naturally expire, and UK/EU law offers limited guidance on post-termination exploitation of identity. In transactions of this nature contracts must therefore grapple with:
- post-termination use of trained models;
- exploitation after the individual’s death;
- retraining using historic data; and
- sunset periods versus perpetual licences.
Closing thoughts
Lame’s value lies not in a catalogue of copyrighted works, but in his persona being cross-cultural, language-agnostic and instantly recognisable — making him a particularly strong candidate for AI-driven replication and localisation. While this transaction may be exceptional in scale, it is unlikely to remain unique.
As generative AI matures, practitioners should expect to see more transactions in which individuals license not just their content, but their commercially exploitable presence and identity. For transactional IP lawyers, this represents a shift away from documenting the exploitation of recognised rights, and towards constructing and drafting for new asset classes through contract in areas where the law has yet to catch up.