We can no longer know if a video is real or not

In 2018, when the first Deepfakes began to circulate, we said that this was the beginning of the end of the credibility of the video as proof. We are wrong. It was not the beginning of the end. It was the first warning that the end was close.

Now, with I see 3 from Google, that moment has arrived. The clips that circulate through Reddit and X They are indistinguishable from real recordings. We do not talk hands with six fingers or faces of the disturbing valley. We talk about videos that exceed any visual test that we could do as simple spectators. And if they don’t succeed, they will achieve it shortly. Especially in a world that consumes video on small mobile screens, less retailers.

The end of video as an irrefutable proof

For decades, the video has been our gold standard for evidence. “I have seen it with my own eyes” was synonymous with total certainty. A recording was enough to knock down governments or demonstrate innocences. That axiom of information consumption is dying before our eyes.

Users of I see 3 They share false news clips about disasters, deceased politicians and violence that never happened. With synchronized dialogues, realistic and physical effects. The model even adds conversations that were not in the Prompt Original, as if it had its own editorial criteria. It is with narrative instinct.

But here is the real problem: it is not just that we can create convincing false videos. Is that we are running out of ways to distinguish the real ones from synthetic without sophisticated technical tools. We are entering an era of permanent visual agnosticismwhere each video will imply the question: “Is this really happened?”

The perfect alibi to deny reality

Macron was pushed by his wife when he got off the plane. The Elyseo denied that the video was real: “It is not a real video, it is made with AI.” He had to end up admitting that it was real, but the damage was done: they had discovered the perfect alibi.

If any video can be synthetic, then any awkward video can be discredited simply claiming that it is artificial. The AI ​​becomes the universal escape argument. The politician caught, the company violating rights, the regime documenting repression … all have now The most perfect reasonable doubt ever invented.

It is not necessary to demonstrate that a video is false, just sow suspicion. In a world where falsification is technically possible, the possibility becomes sufficient argument.

Paradoxically, a technology that allows us to create perfect fiction also allows us to deny the perfectly documented reality.

Learn to live without visual certainties

Google’s safeguards are selective: you can’t generate Biden falling, but natural disasters and urban violence. They protect you from the obvious, not from the subtle.

If each video can be false …

• What happens to a society that bases your knowledge of the world on audiovisual consumption?
• How do we judge credibility, guilt or legitimacy when any test can be manufactured in minutes?

The solution cannot only be technical. We need media literacy that assumes counterfeiting as a starting point. But above all, accept that we have lost one of our most basic instruments to distinguish reality from fiction.

The video as proof has died. Or it is about to do it. We only have to learn to live in a world where Seeing Is No Longer Believing.

In Xataka | How to try I see 3 to create videos with Gemini: Requirements and steps to use it

Outstanding image | Google Deepmind