High five: British technology companies set the pace in AI

Last year America’s actors went on strike over pay and the implications of artificial intelligence increasing automation of their roles. The former issue was very much in the here and now; the latter was all about the future, one that stretches far beyond Hollywood and the movie business.
Traditional structures throughout the creative industries are being challenged by a technological revolution. Generative AI allows machines to create new content from prompts. Fresh writing, music and images can be churned out at the click of a button. It has put Britain’s creative and technology sectors, two cornerstones of the economy, at loggerheads.
This seemingly magical content does not appear out of nowhere. It is based on enormous amounts of material generated by human beings. Since the quality of this input determines the output, developers want to train their AI on the best available content. Many in the creative sector believe that technology companies have been breaching copyright on an industrial scale, “data scraping” web publishers to improve their products.
Lawsuits have begun, such as that between The New York Times, the publisher, and OpenAI, the developer of ChatGPT. Licensing deals are being struck between media companies and technology groups. New regulations and guardrails are likely to emerge.
The growth of generative AI also has profound implications for human creativity and jobs. Amid the tumult, promising technology companies are poised to capitalise. According to fundraising data compiled by Beauhurst, the research provider, these are five of Britain’s best-backed private companies aiming their technology at the creative industries.
Synthesia
Wiping out the need for professional actors, studios, sound or camera crew, this “synthetic media” company creates AI-generated videos hosted by unerringly lifelike AI avatars that can speak in multiple languages. Its high-quality, low-cost films are aimed at corporate clients for training, marketing and communications.
Founded in 2017 by Victor Riparbelli, Steffen Tjerrild, Matt Niessner and Lourdes Agapito, it is backed by the deep pockets of Accel, Nvidia and Kleiner Perkins and has raised about £115 million to date. According to Riparbelli: “Our customers tell us that the content they’re making with Synthesia would never have been a video if the company didn’t exist. It would have been a text document or a PowerPoint slide.”
While the company does not aim to create Hollywood films, Riparbelli believes that soon it will make adverts that can then be made in hundreds of variations. It has already made David Beckham appear to speak nine languages to raise awareness of malaria.
“There’s still so much to go on the core technology in terms of quality and fidelity. The total addressable market is constrained by how good the avatars are, how capable they are of doing things.”
ElevenLabs
Mati Staniszewski and Piotr Dabkowski co-founded ElevenLabs in 2022, inspired by watching terrible dubbing of hit American films in their native Poland and determined to do something about it.
The company quickly gained attention for its advancements in AI audio technology, focusing on creating highly realistic and natural-sounding synthetic voices. It has created a platform that allows users to convert written text into natural-sounding speech that can mimic human intonation and emotion. Sequoia and Andreessen Horowitz, the Silicon Valley investment stalwarts, are among its backers, giving it a total of £79.5 million and making it one of London’s so-called unicorns, a company valued at above $1 billion.
The technology is used for audiobooks, virtual assistants and content creation. Audio can be translated into any tongue, wiping out language barriers. The strapline reads: “Ensuring content is understood by everyone, everywhere.”
However, it raises big ethical and risk questions. Voice cloning is already being deployed in extortion attempts and misinformation campaigns. ElevenLabs’ terms ban people from using the platform to impersonate or harm others. Five months ago it introduced no-go voices that “restricted the creation of voice clones that approximate the voices of political figures, including those involved in presidential elections in the US and UK”.
Stability AI
Stability AI has been hailed as Britain’s answer to America’s generative AI companies. Established in 2019, its Stable Diffusion product can instantly create artwork from text commands.
It has had a rocky period. Emad Mostaque, its founder, left under pressure this year and it now has a new chief executive, Prem Akkaraju, and new investment to try to get back on track, bringing its total financial backing to £196 million. It, too, is caught in the middle of the AI copyright debate. A court case continues with Getty Images in Britain and the United States over the alleged use of their pictures within its technology.
“People can anticipate even more powerful models from Stability AI. We see huge potential for our models to increasingly be used across the creative industries from content creators to major studios to games developers”, a spokesman for the company said. “By aiding experimentation, speeding up workflows and providing brand new tools for this community, we will enable them to really push the boundaries of human creativity.”
Signal AI
Signal AI started life in a garage more than a decade ago. Its founders David Benigson, the company’s chief executive, and Miguel Martinez, the chief data scientist, drew in potential investors with the offer of free MOTs at the car repair business next door.
Billing itself as “disrupting market intelligence”, today it counts Hearst and Guardian Media Group among its backers, having raised £77 million. London is its base, with offices in New York, Hong Kong and Lisbon. The company uses AI to analyse and provide insights from vast amounts of “unstructured” data. Its customers include Meta, HSBC, Google and Pfizer.
“It is not the technology that’s going to automate or take large parts of the workforce’s roles,” Benigson said, “but it’s people who know how to use the technology who might do that. These tools have the capacity to supercharge us and scale us. We need to know how to use and apply them.”
Under the new government, he would like to see the issue of intellectual property tackled, “where copyright is protected but there are updated laws and frameworks to clarify how to use this data for AI training models”.
SafeToNet
Giving your child a mobile phone without safety software is like giving someone a car without a seatbelt, according to Richard Pursey. He founded SafeToNet with his wife, Sharon, horrified at the unfettered access young people had to harmful content. “I can remember the feeling of anger that children are growing up around the world and they’re seeing stuff that their parents haven’t got a clue about.”
The company believes that parental controls as they stand do not work. It has built an AI technology called harm block that stops harmful content appearing on devices and prevents children from taking explicit pictures of themselves to share (a leading source of child sexual abuse material).
While the business broadly supports Britain’s Online Safety Bill, Pursey believes the government made a mistake in targeting content providers and argues that safeguards should be embedded into handsets. The company is in discussions with one of the world’s largest manufacturers to do exactly that.
Employing 200 people in Canada, Britain and Germany, where it bought a small mobile phone business, it has raised £42.6 million, mainly from private investors. Pursey said his team had not gone to financial institutions to raise money because they were too interested in financial returns. “The by-product of safeguarding brilliantly is wealth creation, but that’s totally secondary.”