Originally appeared on Medium | Written by Dan Neely Co-Founder and CEO of Vermillio
Note: This paper was written and shared privately in 2020 with my trusted network. Which led to the formation of Vermillio
15 years ago, I was working on the idea for my new company: a software business that used social data to understand consumer wants, desires and needs. While this is an obvious business plan today, numerous Fortune 100 executives told me that people would never contribute their thoughts and feelings to the internet; in fact, many told me that blogs and forums were only for gamers and adult content.
Today [author’s note: this white paper was written in early 2020], people are saying the exact same things about deep-fakes and other generative content. Many are ill-prepared for the speed of change that is happening in terms of digital assets and content.
Digitization has transformed the ways that creative work is made, promoted, distributed, and sold. Industries have been changed beyond recognition. But technology-driven transformation is far from over for people who make a living from images (still and moving), sound, or the written word.
We must now prepare for a new stranger in town — the emergence of synthetic content and Generative AI.
Executive Summary
1. Generative content based on training data derived from existing digital content has arrived (training data are a set of facts, such as an observed video online, from which a machine learns). The question for IP owners is whether they will enable this change with a new royalty model (Spotify) or simply put their heads in the sand (surrendering to Napster).
2. The digitization over the last 10 years of visual, audio and written content leaves these assets unprotected by traditional intellectual property (IP) regulations.
3. Although this seems like a threat at first glance, there is in fact an opportunity for huge long- run upside for IP owners, who can now further monetize their IP and enable deeper consumer engagement while licensing more obscure content. IP owners will create new revenue streams by enabling the creation of synthetic content.
4. There is also the potential for immediate benefit by making today’s licensing processes more efficient and technology enabled.
What’s synthetic content?
Synthetic content is a catch-all term for media created by machines using generative AI and training data we call the digital signature. The digital signature is simply the 1’s and 0’s that represents the content and what a computer reads at an atomic level. The digital signature comes from existing IP such as a movie, artwork, voice, or song, or a wholly new piece of creative made by artificial intelligence (AI) based on patterns in existing datasets.
Examples of existing synthetic content include:
- A personalized video message from a synthetic version of a celebrity (a digital twin)
- A faked voicemail from an executive ordering a financial transaction built with the digital signature of their voice
- A starring role for long-dead actors in a new Hollywood movie using synthetic versions of their digital signature (image and voice)
- A newspaper article created by a machine using AI trained with existing published media content
- AI creating video games using training data but no code
- A fashion model created by AI from the combined digital signatures of active and retired supermodels
- Personalized corporate videos, built with AI
In each of these examples, training data is utilized by AI and other emerging technologies to create something machine-made that has the appearance and persuasiveness of content made by people.
Why does synthetic content and generative AI matter?
We are already seeing an explosion in synthetic content based on the existing digital signature of creative content.
Studios, music labels, gaming companies, actors, singers, brands and business leaders made a conscious choice when they went from analog formats to digital (such as streaming) because of the massive benefits and efficiency of distribution. But they also made an unconscious choice which could greatly impact their futures.
The big news is that there is no legal protection for studios and other IP owners where it comes to synthetic content based on the digital signature. That is because the 1s and 0s that describe part of a Jay-Z song in digital form can be used to create synthetic content where the artist appears to be performing any song. This new creation is neither a mash-up of actual Jay-Z recordings, nor an infringement of his rights to his likeness — it’s an entirely new creation based on a small sample of his digital footprint. There is no federal law in the United States that provides protection to “data as an asset”. This means that third parties can take a digital signature and do what they like with it.
This reality presents the creative industries with a choice. With no room for legal recourse to assume control of synthetic content created from the digital signature of a movie clip, song or image, IP owners must either
- Embrace an emerging status quo where unauthorized third parties will use synthetic content as they wish, paying nothing for it;
- Adopt a new platform where existing technologies including AI and blockchain are used to allow third parties to buy access to an authorized digital signature of training data that is digitally controlled by an IP owner.
While it is likely that some will initially recommend option 1, arguing that the creative industries should lobby for new laws to protect the digital signature, the reality is that securing legal protection is a long and difficult journey with no guarantee of success. The crafting of an effective law would be very challenging; and tiny tweaks to any individual digital signature might allow creators to argue that their work is new.
At the same time, there are massive prospective economic advantages to option 2 that will ultimately make that route far more attractive to incumbents.
This story sounds familiar…
This shift in industry model is reminiscent of the journey the music industry started on with its transition from CDs to streaming music. First came Napster, and eventually Spotify. Streaming services have demonstrated that most consumers would rather pay $10 a month for the convenience of unlimited, personalized music over free (but pirated) alternatives. The timeline between Napster and Spotify was almost a decade mostly due to the fact that the music industry had strong legal footing that protected their IP.
The music industry pivoted to capture some of the benefits for artists, most of which the labels were not set up to do business with due to scale — a positive outcome that came as a surprise following the earlier collapse in sales of recorded music. In fact, many little-known artists who would never have secured a record deal have benefited, finding a streaming audience for their work — and some have gone on to become stars thanks to the distribution power of platforms such as YouTube.
Now the creative industries need to recognize that all of the content they own is open to being repurposed by third parties using a digital signature. Without any of the protections afforded to the music labels, we believe that the winners will be those who embrace this major disruption, enabling a platform that makes it easy for everyone to access a synthetic but approved version of the content with which they want to create something new.
The opportunity for IP holders in a world of synthetic content and Generative AI
A similar story can play out for generative AI.
We are building a platform where artists, studios, IP owners, public figures and even private individuals place an authorized digital signature for all of the IP they own, including their own image, voice or written words.
[Author note — this company now exists — visit www.vermill.io]
Third parties of all kinds — companies, individual citizens, public bodies, non-profits — pay for the opportunity to access the authorized digital signature. Payments would be handled by the platform. This same platform can be used to manage the IP transactions currently occurring at studios, labels, gaming companies and others. However, this approach does not need armies of lawyers and style sheet experts, as it is managed using smart contracts.
For the owners of catalogues of music, movies, writing and images, there are significant opportunities. For example, most IP is never licensed today because of the administrative burden associated with granting a third party the legal right to use a few seconds of a movie, an article, or an image. Instead of the IP market being based on a few huge deals between, say, a movie studio and a handful of big brands, the Trace platform would create easy access to IP for everyone.
Imagine that a company wants to access a moment in Toy Story, allowing their marketers to create their own synthetic content — for example, creating a unique experience with Buzz Lightyear on TikTok.
The platform provides a toolkit for creators to use the data to ultimately build new content. IP owners would receive payments any time their data is used, while also limiting foul play with their IP in the form of deep-fakes or other forms of unauthorized content. Potential bad outcomes could be prevented by using technology to highlight synthetic content created without the seal of approval of the person or asset featured.
The platform will help creatives, studios, brands, and individual consumers to create new content drawing on a library of training data and the power of immense distributed compute capacity. It will transfer the face of creativity, putting power in the hands of everyone. It’s good news too for established talent. Imagine a version of Cameo that is unconstrained by the time available to the celebrity in a given day because consumers can decide to buy a video from the actual celebrity or from the synthetic version that they control. Imagine a version of the Minions or Dom Torretto wishing you a Happy Father’s Day with a personalized greeting, and the studio getting paid each time that happens.
Furthermore, brands and others will be able to create entirely new synthetic content — written, audio or visual — using AI, such as a new car in Fast and Furious composed of the three most epic cars of all time, or a new bad guy derived from the three most evil Bond villains. This will accelerate the creative process in video games, MMOs, movie production, AR/VR and more.
Of course, the platform will also allow for engagement, recognizing that the content creation that happens needs to be shared in some capacity. A consumer wants to actually dance with Stuart the Minion, or sing with Beyoncé and post to TikTok. The cost/benefit of a studio using traditional methods to create a custom Stuart for every consumer does not work, and Beyoncé has no interest in singing with every TikTok user.
However, the synthetic digital twin of each of these have all the time in the world and can be created using our platforms, and distributed on the Engagement platform (which could be an owned platform). Our platform will allow for sharing of and engagement with synthetic content.
The emerging world of Generative AI
For the owners of IP, authorized digital signatures made up of training data represent an opportunity to monetize the desire of third parties to create synthetic content, while also highlighting unauthorized synthetics that might be used for ill.
Back catalogues can be opened up for citizen creators and companies looking to access the long-tail of existing content in the same way that eBay created a marketplace for buyers and sellers of every object on earth.
Studios can start to create marketing campaigns that have thousands of derivations personalized to the target consumer. The computer only needs 5 seconds of voice, or 15 seconds of video, to create any type of new content.
A famous actor that defines a franchise will no longer have to come in and do voice edits. That same actor can be recreated over and over again without ever having to step foot in a studio. Fast and Furious 50 starring Vin Diesel anyone or perhaps Minions 110 starting Steve Carell. An actor no longer has to go into the studio for voice-overs. These can either be completely synthetic; or an amateur actor can do the reading and it can be changed to the voice of the star with their digital signature. Estates of deceased actors, singers and writers can now continue to monetize. A synthetic Sinatra singing the latest James Bond theme is now possible.
For example, a new animation franchise can be created using current IP. While it may take three years to develop Minions on the current approach, a new franchise could be created using the digital signature of Minions that looks radically different, but is based on Minions. First, we show the computer the first frame (with Stuart with his mouth open) and then the next 1000 frames (to where Stuart is hugging another Minion). Now the computer can replicate all of this for our new animation franchise based on farm animals. We create the first frame and the last one, but the other 1000 frames would be created by the computer. This would allow inventive artists to spend less time on labor-intensive, frame by frame redaction, so they can invest in additional fascinating things. In this way, as noted above, incumbents can use Generative AI to create even better new work.
Time to act
The creative industries must recognize that a new world has arrived where there is no legal precedent and their data and content will be used. This puts existing business models under threat, just as we saw in the music industry that saw its revenue fall by almost 50%.
The choice — allow this to happen, or instead buy into the platform that can provide authentication, convenience, and the technological capabilities to enable synthetic creativity.