When OpenAI unveiled Sora 2, it was billed as the next big leap in digital creativity, a text-to-video tool capable of conjuring entire cinematic scenes from a single sentence. Meta quickly followed with Vibes, powered by its Emu Video technology. On the surface, these platforms look like the next evolution of storytelling. But beneath the novelty lies a darker, more existential challenge: we are losing control over what it means to be real.
As someone who has literally lived this, I was deepfaked. I can tell you firsthand that this technology doesn’t just imitate identity; it hijacks it. Seeing a synthetic version of myself say and do things I never did was more than unsettling; it was a wake-up call. The deepfake wasn’t perfect, but it was close enough to possibly fool those who didn’t know me personally. That’s the danger. The technology to create hyper-realistic video has outpaced our ability to detect and defend against it.
The “Cameo” Problem: When Your Likeness Becomes Public Property
The most controversial feature of Sora 2 is something called “Cameos.” It allows users to open-source license their likeness, enabling others to insert that identity into AI-generated videos. OpenAI CEO Sam Altman became the first to offer his image publicly, a bold experiment in what he called “shared digital creativity.”
Almost immediately, videos began surfacing of Altman skateboarding, plunging toilets, and performing Jackass-style stunts. Some were funny. Some were malicious. And all were indistinguishable from reality.
This open-licensing experiment introduces an entirely new legal and ethical paradox: plausible deniability. Once you’ve licensed your image to the world, how do you prove, or disprove, that any video of you is real?
Even cybersecurity veteran Perry Carpenter from KnowBe4 highlighted this issue in a demonstration video released the same day Altman made his announcement. His deepfake illustration wasn’t just convincing; it was a blueprint for how easily our identities can be repurposed without consent.
Meta’s Vibes vs. Sora 2: Two Roads to the Same Reality Crisis
While OpenAI’s Sora 2 promotes identity-sharing through verified licensing, Meta’s “Vibes” takes a less invasive route. Using only a selfie and a text prompt, users can place themselves inside AI-generated videos.
The difference lies in control:
- Sora 2 allows other users to feature your likeness with permission — effectively creating a marketplace of identities.
- Meta’s Vibes is designed for self-use only; there’s no system for licensing your face to others…yet.
Meta’s Vibes (Emu Video)
— Likeness Capture: Single selfie
— Use Case: Personal creation
— Identity Verification: None
— Core Concern: General misuse
OpenAI’s Sora 2
— Likeness Capture: Verified video + audio
— Use Case: Others can use your likeness
— Identity Verification: Required for Cameo
— Core Concern: Deepfake licensing & plausible deniability
Sora’s Cameo feature opens the floodgates to synthetic celebrity culture, but also to irreversible loss of control. Once your likeness is uploaded and licensed, the videos made with it can spread across multiple platforms, duplicated endlessly. And once they’re in the wild, you can’t claw them back.
The Next Weapon in the Disinformation War
The more troubling aspect of Sora 2’s evolution is its capacity for synthetic propaganda. Early users have reported the ability to generate hyper-realistic videos depicting children in war zones, a category previously blocked in earlier models. These clips are almost impossible to distinguish from authentic footage.
When manipulated narratives can be fabricated in minutes, and social media remains the dominant news source for millions, the implications are staggering. Disinformation will no longer need a writer; it will have a director.
This isn’t about entertainment anymore. It’s about influence, and potentially, psychological warfare.
Detection Can’t Keep Up — So It’s Time for New Rules
Right now, detection technologies lag years behind the generative tools. Watermarking and metadata tracking are unreliable once videos are downloaded, clipped, or re-uploaded. The current ecosystem offers no real chain of custody for digital truth.
We need to rethink how authenticity is verified, not after a video is viral, but before it ever leaves the source. That means building new controls, powered by the same AI foundations that create these videos, to authenticate reality in real time.
From a cybersecurity standpoint, we might even need a temporary regression in trust models. Going back to something simple and human. For example, in critical situations, organizations could use shared passphrases or challenge-response codes to confirm identity. It’s old-school, but it works. Sometimes the answer to high-tech deception is low-tech verification.
The Legal and Ethical Fallout
The long-term ramifications of open-sourcing identity are profound. Once your likeness is uploaded and used, it can be replicated, transformed, and redistributed infinitely across platforms, borders, and contexts.
This creates a nightmare for courts, law enforcement, and digital forensics. In a world of infinite replicas, how do we prove authenticity? Legal systems will need to adopt a digital chain of custody, much like evidence handling in criminal investigations. Without it, anyone from CEOs to politicians could claim any compromising footage is “AI-generated,” and there would be no way to definitively prove otherwise.
That’s the essence of the coming crisis: plausible unreality.
Spotting the Synthetic: What You Can Still Look For
Even as deepfakes grow more convincing, subtle cues sometimes remain:
- Micro-expressions: Real faces move with tiny, involuntary asymmetries that AI can miss.
- Lighting mismatches: Watch how shadows and reflections behave; they’re rarely perfect.
- Audio sync issues: Slight lags between sound and movement can reveal a fake.
- Context validation: If a clip triggers outrage, pause before sharing. Verify with multiple sources.
- Reverse search: Use forensic tools like InVID or Google Lens to trace origins.
But here’s the truth: detection will never be enough on its own. We must start building trust frameworks, not just trust tools.
Reality Has No Undo Button
In the age of AI-generated video, the line between real and synthetic has evaporated. These technologies democratize creativity, but also democratize deception. Once a fake enters the ecosystem, there’s no delete key for belief.
For those of us who live in both the human and digital worlds, this isn’t just a technical issue; it’s an existential one. We’re witnessing the erosion of visual truth itself.
The only way forward is through transparency, digital chain-of-custody systems, and the courage to admit that even the most advanced AI can’t fix a problem it helped create, unless we teach it to tell the truth.
About the Author
Len Noe is a Solutions Architect at BeyondTrust, Transhuman, Podcaster, International Cyber Security Speaker, Author, Technical Evangelist, and Biohacker with 11 implanted microchips. A former blackhat and ex-1% motorcycle club member with more than 30 years in technology, he has presented in over 70 countries and is featured in the documentary I Am Machine, which premiered at DEF CON 2025.