Deep Real unveils 'medium after video': real people, interactive, no AI or deepfakes

Two dates frame the story. On 30 September the Governor of California must sign or veto SB 1111 and SB 1050 — two bills that write "digital replica" into state law and require adverts built on a synthetic performer to say so. On 1 July SAG-AFTRA's new contract came into force with, in its chief negotiator's words, rules to ensure "synthetics remain the exception in our industry instead of the rule." Both are attempts to police a technology that fakes people. Deep Real does not fake people. It is not an AI company and its product is not a chatbot, an avatar or a clone. It is a new medium, sitting between the two things the internet already has — video you can only watch, and simulations that can only pretend — and it works like this. A person sits down and is filmed, properly, answering the questions people actually ask them: how they trained, what they were afraid of, what they would tell their younger self. Those filmed answers are structured into what the company calls a living story world. A viewer asks a question, in their own words, and the system retrieves the real filmed answer and plays it. The cut between one answer and the next is hidden by a transition the company calls the Invisible Cut — the viewer's own camera image briefly fills the screen as the next segment loads — and that transition is the subject of US Patent 11,483,494, granted on 25 October 2022. OpenAI released ChatGPT on 30 November 2022. The consequences follow from the design. Because every reply is a recording, the person can never be made to say something they did not say. Because nothing is generated, there is no hallucination to guard against and no synthetic performer to disclose. And because the answers are files rather than inferences, they are delivered the way video is delivered — over standard content networks — rather than the way chatbots are, through expensive real-time computation. "Every word and every frame on our platform is real, nothing generated, nothing synthesized," Tim Levy writes in the company's manifesto, Closing the Intimacy Gap. Levy is a 20-year entertainment-industry veteran — his film-finance work included Pirates of the Caribbean 2 and 3 — who now runs Deep Real from Los Angeles and hosts the podcast Echoes Across Time. His argument is that reach has never been greater and genuine access has never felt further away, and that the industry's answer to that gap — synthetic versions of real people — is the wrong one. The evidence of 2026 is on his side. In November 2025 an app that let people talk to AI avatars of dead relatives was called "nightmare fuel" by 22 million viewers of its own launch video. In February Soul Machines, the "digital people" pioneer that raised around $135 million, went into receivership. In March OpenAI announced it was shutting the Sora app, whose likeness feature had drawn objections from Bryan Cranston, SAG-AFTRA and the estates of Martin Luther King Jr and Robin Williams. Meta's celebrity chatbots were caught flirting with users in the stars' names. Every headline in the category has been a backlash, a collapse or a consent patch — an AI voice marketplace here, a likeness-detection tool there. The one path nobody had taken was to stop synthesising the human at all. Deep Real's first product on that path shipped in October 2023: Messi Connect, an app in which fans could put questions to Lionel Messi and get his filmed replies. Its first commercial focus now is WWE's Natalya "Nattie" Neidhart, the most decorated woman in the company's history — seven Guinness World Records, the latest set in June 2026 for the most submission wins by anyone, man or woman, in WWE history — whose invite-only beta for her first 100 fans is taking applications at deepreal.com/nattie. She is exactly the kind of performer the new rules leave out: wrestlers and athletes are not covered by SAG-AFTRA's contract, and the federal NO FAKES Act, which cleared the Senate Judiciary Committee in June, carries an exemption for sports. The format is built for anyone whose value is themselves: athletes, cultural icons, faith leaders, experts, educators, and — the use case the manifesto returns to — families. A parent filmed once, answering the questions their children will one day want to ask, is a very different object from a generated parent. One is a recording. The other is a guess.