Psst… We received a tip…

At 8:59 a.m., your AI work avatar clocks in. Wearing your face, shirt pressed, smile loaded, voice steady. It opens the deck, records the training clip, walks the client through the numbers, and never once forgets the line you meant to say.

By 9:03 p.m., a different version of you takes over. This one is here to look good, change outfits, try on personalities, and drift through the internet like it was born popular.

That is the split hiding inside this week’s story. One company seems to be building the version of you that shows up for work when you are tired, or nowhere near the camera. The other seems to be building the version of you that is the new age socialite.

This week’s patents come from Google LLC and Meta Platforms, and together they read like a tale of two digital selves.

This week’s filings come from Google LLC and Meta Platforms, Inc. 

Google’s Patent

Start with Google. You give the system video of yourself. It uses that visual data to build a personalized avatar, then takes a plain-English instruction like “present the second half of my slide deck” and generates content that makes the avatar carry out the task. 

First comes training, where the model learns from examples that pair instructions with 3D representations of humans doing actions. Then comes supervised fine-tuning, where the system is told what body parts matter most for a given task, like lips for speech or hands for gestures. Then comes human feedback, where a person scores the output and helps the system get better at making the avatar behave believably. 

The patent’s examples lean toward work-like use cases, including presentations, social content, and question answering. It is closer to a digital co-worker.

Meta’s Patent

Meta starts from a different place. Its system takes prompts like “make me a pirate” or “show me in streetwear,” then breaks those requests into smaller choices across hair, clothes, face, body, skin tone, eyewear, and facial hair. 

The social-trends feature is the tell. Meta’s filing says the system can look at social media trends and use that information to suggest styles and accessories. So the avatar is being edited to match the culture around your prompt. Google wants your avatar to perform. Meta wants it to belong.

More of life now has to show up as media, but making that media is still slow, awkward, and expensive. Google is trying to solve the “I need a polished version of me to show up for work” problem. Meta is trying to solve the “I need a version of me that fits the latest trend and the platform” problem.

Being on camera is its own chore. In a 2024 survey of 1,360 Americans, 48% said video calls caused hyper-gaze anxiety and 46% said seeing themselves on screen triggered mirror anxiety.

Google has been pushing Flow as an AI filmmaking tool built on Veo, Imagen, and Gemini. Google now positions Flow as an AI filmmaking tool for creatives, and Google Vids is explicitly sold as an AI-powered video creation app for work that helps people “scale your message through video storytelling,” with prompt-based creation, teleprompter support, AI voiceover, and now AI avatars inside Vids.

Meta has been pushing AI Studio, including creator AIs built as an extension of the creator, which is a direct bet that social presence now needs to scale beyond the human behind the screen.

On the work side, startups like Synthesia are already selling AI presenters to businesses that want training videos, demos, and explainers without full production overhead.

On the social side, companies like Genies and platforms like Roblox are already selling the idea that digital identity is something people will build, wear, and spend around. (Reuters)

On the work side, companies want faster ways to make training videos, demos, onboarding clips, and sales content. Vimeo says 65% of organizations have experienced a surge in video content creation in the past two years, 73% expect volume to keep accelerating, and video is now “essential” for learning and development, HR onboarding, sales prospecting, remote product demos, and customer support. Vimeo also says 48% of organizations struggle to scale video creation while maintaining consistent quality. 

On the social side, people already spend real time and money building digital selves across apps, games, and virtual spaces. Snap says Bitmoji is used by over 320 million people globally every day across profiles, Snap Map, AR Lenses, chat, text, and email. Roblox says users are increasingly purchasing avatar items like emotes to express personality, and 73% of surveyed Gen Z Roblox users want their avatar to express a full range of emotions.

Goldman Sachs estimated the creator economy at about $250 billion in 2023 and said it could reach roughly $480 billion by 2027. (Reuters)

Meta AI reached 1 billion monthly active users by late 2025, while Roblox reported 111.8 million daily active users in the second quarter of 2025 and said creators had earned over $1 billion in the prior year. The audience for digital workers and digital selves is already enormous. (Reuters)

Companies turning you into an AI video or avatar are already raking in investment. Synthesia, which lets businesses make corporate and instructional videos with AI presenters, raised $180 million at a $2.1 billion valuation in 2025, then another $200 million at a $4 billion valuation in 2026. 

Higgsfield, which builds AI video tools aimed heavily at marketing teams, raised $80 million in 2026 at more than a $1.3 billion valuation. Higgsfield’s ad tool says it can turn a product URL into a video ad with “AI avatars and voiceovers,” and it frames itself as infrastructure for creators, agencies, and marketers producing video and visuals at scale. 

Hedra, which helps brands make lifelike digital characters and spokesperson-style videos, raised $32 million in 2025. (Reuters)

The social-avatar side is getting funded too, but the pattern is a little different. Genies, which builds digital-avatar tools and ecosystems for creators, brands, and talent, hit a $1 billion valuation with a $150 million round.

Inworld, which builds AI character tools for games and interactive experiences, pushed past $100 million in total funding at a valuation above $500 million.

Ready Player Me, which lets users create avatars that move across games and virtual worlds, was acquired by Netflix in late 2025. Investors and buyers still care about digital identity, but they seem most interested when avatars become infrastructure for games, creators, or IP. (TechCrunch)

The clearest legal risk is misuse of likeness. The moment a system can generate or edit a convincing digital version of a person, you are in publicity-rights, privacy, fraud, and consumer-protection territory. That risk is sharper for Google because its avatar is meant to perform tasks and speak on cue, but it is still very real for Meta because fast avatar editing can slide into impersonation, deceptive endorsements, or brand confusion if the line between stylized self and synthetic replica gets blurry.

Regulators are already moving. The FTC finalized its impersonation rule for government and business impersonation in 2024 and separately proposed extending protections to impersonation of individuals. In Europe, the AI Act’s transparency rules for AI-generated content are due to apply from 2 August 2026.

Google is trying to build the version of you that can show up to work, hit its marks, and keep the machine moving. Meta is trying to build the version of you that knows how to embody you and be popular online.

They are early claims on two different futures for your second self.

If you want to read the source for yourself, start with Google’s US 2026/0045018 A1 and Meta’s US 2026/0087707 A1. Then ask yourself which version of you Big Tech wants first.