Navigating the Future of AI in Voice Acting Contracts: Insights from Industry Experts

10 July 2025

Navigating the Future of AI in Voice Acting Contracts: Insights from Industry Experts

What's in the fine print? As the role of AI in the localisation workflow continues to evolve and AI-generated voices become increasingly more sophisticated, the lines between human performance and machine synthesis have begun to blur. This shift isn't purely technological; it's contractual, ethical and deeply personal for the artists behind the mic – bringing into focus the importance of consent, control and compensation. This is also not a straightforward, two-sided conversation between entertainment creators and voice talent, as several other stakeholders are in the mix.

At an event in Los Angeles, VSI brought together a panel of voices from every corner of the conversation – a studio, LSP, AI developer, voice actors and active SAG-AFTRA and NAVA members – to have the kind of open, honest dialogue this topic demands. The goal wasn't to "settle" the debate but to broaden the voices at the collective table, approaching with transparency, nuance and a willingness to confront what most avoid because it feels uncomfortable, laying the groundwork for a more informed, collaborative future and leaving everyone in the room with one essential question: What can I do to keep the conversation going?

Assignment of Rights and AI

As technology evolves, so too must the respective legislation. As an industry, we've seen this impact on the traditional assignment of rights becoming more complex within the dubbing contract. William Russell, Deputy CEO and former General Counsel at VSI in London, explained: "The debate around the transfer of talent rights as it concerns AI starts with the assignment of rights agreement, which is a contract that assigns the talent’s IP in the work to the dubbing studio. Prior to the introduction of AI, traditional AoRs were widely drafted and included phases such as the right to edit, alter, modify and replicate, which were required for editing purposes, such as splicing lines or adjusting timings to fit lip-sync. However, when now viewed in the context of AI dubbing, concerns arose that this wording could be interpreted as assigning the right to exploit talent IP for reasons that weren’t originally intended, or that the talent weren’t willing to consent. This highlighted the need for greater transparency to protect performers in the age of AI.”

Union Action Across Borders

Unions are playing an increasing role in negotiating AI usage in contracts, focusing on the issues of consent, control and compensation. As the technology continues to advance, it brings into question the rights around character simulation and digital double rights. This was one of the contributing factors to the SAG-AFTRA actors' strikes that we saw in 2023 and beyond. Zeke Alton, an actor and volunteer negotiator for SAG-AFTRA, shared insight into the union's efforts: "Before the TV/theatrical negotiations, SAG-AFTRA was engaged in commercial negotiations where they tried to bargain the use of performer services for AI replication, digital double rights. As this technology hit the forefront of the news, we then saw strikes in TV, film and interactive contracts. Since then, we've been working in partnership and bargaining to work our way through and come up with proper consent, compensation and transparency.”

While much of the highly publicised union activity around AI in voice acting has been concentrated in the US through SAG-AFTRA, Europe has seen its own reaction, most notably through PASAVE and United Voice Artists (UVA). PASAVE (Plataforma de Asociaciones y Sindicatos de Artistas de Voz de España) emerged in Spain in response to the growing use of AI in dubbing and voiceover. As Tim Friedlander, President and Co-Founder of the National Association of Voice Actors (NAVA) detailed, "PASAVE launched specifically to address AI and became a central voice in the European conversation. The group initially proposed a single AI clause.” But what began as a unified front soon evolved in its complexity. Tim explained, "That single clause now turned into four variations, depending on whether companies accepted it outright, modified it or rejected it entirely.”

This fragmentation highlights the difficulty of achieving consensus across a diverse industry landscape. At the content owners' level, Scott McCarthy, VP of Localization at DreamWorks Animation, shared the perspective of how legal complexities restricted opportunities for decision-makers to meet and talk: "Because of EU competition law, the major studios and streaming platforms couldn't be in the same rooms negotiating together in response to what the dubbing unions and associations were advocating for. So, the parties ended up having to negotiate independently, which resulted in a lack of uniform language across the board." Localisation service providers, bound by master service agreements (MSA) with these content owners, often lacked the authority to adopt new AI clauses unilaterally, which was seemingly the expectation from some of the global unions. As William noted, "Dubbing studios do not have an autonomous right to insert a new AI clause without consent and agreement from the end client." This created a bottleneck where contractual realities clashed with ethical intentions.

Global Impact and Legislation

Creating unified legislation across different regions remains a significant challenge. Tim highlighted the varying levels of protection in the US alone, which adds to the complexity: "The US has some of the weakest biometric protection in any country in the world, and only 20% of the voiceover industry is unionised. Currently, on a federal level, no one owns their voice, image, name or likeness. On the state level, there are certain areas where we do. California, for example, has stronger laws along with Washington and Illinois, but depending on where in the country you are located and what location you are working with, at home or abroad, the question remains: what laws are you actually being protected under?"

To help unify and amplify these efforts, United Voice Artists (UVA) was co-founded by Tim and others as a global coalition of voice actor associations. UVA adopted PASAVE's foundational clause and worked to disseminate it across other regions, thereby taking steps towards creating a shared framework for the use of AI. "We adopted what PASAVE had taken and spread that around through the rest of the associations," Tim said. Together, PASAVE and UVA represent a growing international movement that recognises the need for both local action and global solidarity in the face of rapidly advancing AI technologies.

Future of AI in Voice Acting

The future application of AI in voice acting involves both technical and ethical considerations. Andrea Ballista, Co-Founder of VOISEED, emphasised the need for ethical agreements and consent when creating AI solutions: "Part of AI is defining the rules of how information can be managed, and the data source is essential." He shared his candid perspective as an AI developer, highlighting the delicate balance between technological advancement and the responsible use of data. "When I talk about AI, it means data. And the source of that data is essential." He outlined three primary data sources used in AI voice training: proprietary datasets created in-house, existing datasets that can be refined and the vast, often unregulated pool of data available online. While general training of speech models may rely on broad datasets, Andrea drew a clear line when it comes to cloning a specific human voice, which he said "requires a kind of ethical agreement between all the stakeholders." At the heart of this agreement are the now-familiar principles of consent, control and compensation.

It is also essential to consider the realistic uses for cloned voices, both now and in the future, with a clear distinction between the different types of AI voice applications. "Using data to train a general model is one thing. Using it to clone a voice, modify another actor's performance or generate unscripted content is something else entirely." Each use case demands its own set of ethical and contractual considerations. To navigate this complexity, the industry must consider developing shared terminology and technical standards. Not all content demands the same level of human nuance or production value. Andrea proposed the need for a "fit-for-purpose" approach to AI voice integration. "If you're using a free translation model, you don't expect top-tier quality; it's functional and it just needs to work. The same logic can apply to AI voices." In this vision, high-end productions may continue to rely on human talent, while AI could support lower-tier or unscripted content, provided all stakeholders agree on the terms of use.

But Andrea was clear: the future of AI in voice acting isn't predetermined. It will be shaped not just by what is technically possible but also by the collective choices of developers, studios, service providers and performers alike.

Educational Efforts and Industry Collaboration

As AI technologies become more embedded in entertainment workflows, education has emerged as a critical tool for empowering voice actors. Zeke addressed the emotional and informational gaps many performers face. "There are actors who want AI banned entirely and companies who ignore ethics altogether, but most of us are in the middle." That middle ground, where ethical use, consent and compensation can be negotiated, is often overshadowed by fear, especially when performers encounter unfamiliar legal terminology in contracts for the first time. "That fear has to be met with education," he advised, highlighting efforts by SAG-AFTRA and NAVA to deliver complex information in accessible ways. Even the use of an opt-in, opt-out mechanism within legislation has caused confusion, making something seemingly straightforward more complex. Since there isn't a global agreement on its definition, there have been instances where both opting in and opting out were actually referring to the same thing from one country to the next. Tim added that many voice actors also joined the AI conversation late, unaware that their performances had already become "data". This shift in understanding – from voice as art to voice as data – has been jarring for many, and legacy contracts have only compounded the confusion. This ties into a point Scott made earlier in the panel, “lumping all AI under one umbrella dilutes the conversation. Differentiating between automation, voice cloning and generative AI is essential to building meaningful, forward-looking dialogue.”

Consumer Preferences and Ethical AI

While much of the conversation focuses on contracts, technology and talent, a critical voice is often underrepresented: the consumer. Scott pointed out that audience sentiment may ultimately shape the future of AI in the entertainment industry. "I don't think we're giving enough credibility to what the consumer will want. If two blockbuster films open – one made with AI and one without – I tend to believe the consumer will lean toward authenticity." This growing demand for transparency is already visible in other creative sectors, from backlash over AI-generated images to calls for disclaimers in major productions. Andrea echoed this, noting that voice remains one of the last emotional anchors in digital experiences. "Users don't want to break the suspension of disbelief," he said, reinforcing the importance of aligning AI use with audience expectations.

Tim and Zeke added that consumer sentiment is not just theoretical; it is a tangible reality. It's vocal and immediate. From social media uproar over AI-generated game dialogue to public outrage at AI-generated ads, the message is clear: most audiences care how content is made. "There's a perception that using AI means your company is cutting corners," Tim explained. William raised a deeper question: even if AI-generated voices sound authentic, will consumers accept them if they're not ethically produced? As the industry navigates this evolving landscape, one thing is sure: consumer trust will be shaped not just by how content sounds but by how transparently and responsibly it's made.

A Look to the Future

The entertainment industry stands at a pivotal crossroads when it comes to leveraging technological innovation and determining the realistic and ethical application of AI within both the current and future landscapes. Through open dialogue, ethical frameworks and cross-border collaboration, stakeholders can collectively shape a future where AI enhances creativity without compromising the rights and livelihoods of human performers. But one thing is clear: to do so, we must encourage this collective dialogue and remove the fear around having difficult conversations.

The content of this article is provided for informational purposes only. The views, opinions, and statements expressed by quoted individuals or sources are solely their own and do not necessarily reflect the views, opinions, or positions of VSI.