Deepfake Transparency and Identity Rights Get Boosts from Truepic and Metaphysic

New legal and business practices will accompany deepfake's rise

Apr 06, 2023

Deepfake and digital twin technologies pose several novel challenges. One concern is that deepfake quality is getting so high that viewers increasingly face difficulty identifying real from synthetic media. This will become even harder as more of our media mixies the two formats.

Truepic introduced a new cryptographic signing solution this week that offers transparency and provenance information about synthetic media such as deepfakes. In other words, the solution is about who created the media and who owns it.

Also, this week, Tom Graham used the good old-fashioned U.S. legal system to assert ownership over his digital avatar likeness. Graham is the CEO of Metaphysic. The company announced that Graham became the first person to submit and receive copyright registration for his AI likeness with the U.S. Copyright Office.

Both of these use cases are about authenticity and ownership with the latter providing protection for the individual from unauthorized use of their likeness.

Deep Provenance

A deepfake version of Nina Schick was featured in a video created by Revel.ai that included a cryptographically signed authentication from Truepic. This video included a disclosure that it “Contains AI-generated content,” which is visible in the image at the top of this post. It also includes information about when it was created, who created it, and the authenticated signature.

Schick told me in an interview that deepfake adoption has exceeded her expectations since she first started researching the technology in 2017 and even since her book on the topic was published in 2020.

This revolution is even bigger than what I first imagined in 2017, and even then I understood it was a big deal…guided by my background in disinfo and information warfare. The questions I raised in my book around information integrity are more important than ever, hence the focus on provenance.
There [is a big] difference between a simple watermark and the cryptographic hash used to sign this video, which is designed to be interoperable across the internet [and follow the] open standard for media provenance and authenticity being developed by the C2PA.
The idea of authenticity and provenance applies to both synthetic and authentic media. It's about transparency to ensure digital trust.

Truepic’s solution is a record of media provenance. Its Display product allows you to track changes and confirm the authenticity of a media artifact, including a detailed account of an image's journey from a verified source to each subsequent edit. Details such as whether the image was captured by a camera, manipulated in Photoshop, or compressed for display on your device, are recorded in the artifact’s history and available for any user to review.

Deep Ownership

While Truepic is focused on authentication and transparency, Metaphysic’s Graham is focused on protecting individuals’ rights to their likeness and their hyper-real digital representation. Graham said about the step:

"Generative AI can create content that looks and feels real, and regular people's avatars can be inserted into content by third parties without their consent. This is not right, and we should never lose control over our identity, privacy or biometric data."

According to Axios, Graham “became the first person to receive U.S. copyright registration on an AI likeness of himself.”

To produce his AI replica, Graham recorded a three-minute video with his mobile phone, capturing his appearance, voice, and biometric information. Metaphysic then used its deepfake AI technology to generate Graham's AI avatar. The team also incorporated the AI model output into the underlying video by compositing and merging it to create a precise representation of Graham.

Co-founded by Chris Ume and Graham in 2021, Metaphysic is one of the most recognized names in deepfake technology providers. Ume became widely known for his deepfake prowess with his Deep Tom Cruise collaborations for Instagram. Metaphysic rocketed to international attention in 2022 when its technology was featured in the preliminary, semi-final, and final rounds of America’s Got Talent.

Deep Identity

Metaphysic’s first products provided custom deepfake solutions for customers. However, it was not a self-service solution. Trained Metaphysic staff executed the project using in-house, proprietary software. The company recently introduced a self-service avatar builder “for the metaverse” and other online use cases. The coming soon section says you can “Own and manage your digital self using our platform.” This is the first step in managing your ownership rights for your digital likeness.

You can imagine this coupled with a service or how-to guide that facilitates registration with the U.S. Copyright Office or similar authorities in other countries. Metaphysic may enable you to use your avatar in metaverse virtual worlds and with other online services. It could also help you secure ownership over your digital likeness. With the slowing of metaverse adoption, this digital identity protection solution could be a scalable product that completely transforms Metaphysic’s market position and business model.

The global identity theft protection market was valued at over $10 billion in 2022. Digital rights management is another multi-billion dollar market segment. Digital likeness ownership protection and authentication represent new products poised for growth and Metaphysic appears to be laying the groundwork to assert leadership in the space.

Deepfake Utilities

Deepfake and generative AI solutions such as voice clones, digital twins, celebrity chatbots, and virtual humans offer new benefits but also introduce new challenges. Synthetic media use cases have previously focused on how to employ the technology. There is also a call for easy-to-use and reliable solutions to detect AI-generated media to combat fraud and disinformation.

Solutions to manage and protect synthetic media assets will soon become critical infrastructure supporting this growing market. Some examples include authentication, disclosure, tracking, ownership validation, and identity protection. The emergence of this tooling will help drive the adoption of synthetic media and help to mitigate risks.

What do you think? What other use cases do you think are needed? Which will you sign up for first?