4dauth

Video authenticity verification for creators. Every card is unique. Not AI-generated. Not AI detection. Just visual proof.

[4dauth @4DAuth]

4DAUTH.
Video authenticity, verified.

Proof. Not detection.

https://t.co/8yPJWW66BM

[4dauth @4DAuth]

YouTube now detects deepfakes of creators and lets them request removal.
That's the platform doing its job.
What it doesn't do: tell a viewer, right now, while watching a video, that the person on screen is actually the real creator.
Detection is reactive. Proof is what viewers need in the moment.
Proof. Not detection.

[4dauth @4DAuth]

The PR industry is waking up to something creators already know: by the time a deepfake is identified and a response is drafted, the damage is done.
Crisis communications assumes there's a window to respond in. Deepfakes collapse that window.
The only thing that beats a deepfake at publication speed is proof that was already in place when the video was posted.
Proof. Not detection.

[4dauth @4DAuth]

The TAKE IT DOWN Act gives platforms 48 hours to remove a deepfake once reported.

That is a meaningful step. It is also a framework that starts with the fake already published, already seen, already spread.

It does not help a creator prove a specific video is genuine before any report is filed. The law answers "take this down." It does not answer "is this real."

Takedown on demand and creator-published proof are solving different problems on different timelines.

Proof. Not detection.

https://t.co/H7RNyakuNS

[4dauth @4DAuth]

The deepfake threat to creators is not that a fake video of you exists somewhere online.

It is that a version of you is actively directing your audience to hand over money and data. Your face. Your voice. Your credibility. Pointing your real followers at a fake investment, a fake exclusive group, a fake charitable appeal.

Your audience cannot tell. That is the whole point.

Creator-published proof at the source is the layer that draws the line. Before a viewer clicks. Before a follower sends.

Proof. Not detection.

https://t.co/H7RNyajWYk

[4dauth @4DAuth]

Deepfake regulation is now in force in dozens of jurisdictions. The rules don't agree.

What counts as consent in California isn't what counts in the EU. Voice-cloning rules diverge from likeness rules diverge from synthetic-media rules. Penalties, takedown windows, even definitions of "AI-generated" all vary by country.

Creators publish to a global audience. The threat scales globally. The protection won't.

Creator-published proof of origin doesn't wait for one jurisdiction to agree with the next. A viewer anywhere can verify the same way.

Proof. Not detection.

https://t.co/H7RNyajWYk

[4dauth @4DAuth]

How a 4DAuth verification card works, in 30 seconds.

1. The creator generates a card from a seed phrase tied to their handle.

2. They publish the card alongside the video. That pairing is the attestation. One card, one video.

3. The viewer clicks once and sees whether the card, the handle, and the video all match.

The card cannot be lifted and reused on another upload. The seed phrase belongs to the creator, not the platform.

Not a watermark. Not a detector. A proof the creator publishes at the source.

Proof. Not detection.

https://t.co/H7RNyakuNS

[4dauth @4DAuth]

Camera makers are signing video at capture. Sony, Google Pixel, Samsung Galaxy. Real progress at the start of the chain.

But platforms re-encode everything on upload, and when they do, the embedded credential gets stripped. Fewer than 1% of news media on the web today reaches viewers with intact provenance.

Sign-at-capture is one layer. Proof-at-publish is the layer that survives distribution.

Proof. Not detection.

https://t.co/H7RNyakuNS

[4dauth @4DAuth]

Voice deepfake attacks rose 680% year over year. The defenses people keep adding aren't keeping up.

Creators get caught in the middle. Their voice is the easy thing to clone, and a viewer scrolling past has no way to tell which video is real.

The fix isn't a smarter detector at the platform. It's a proof the creator publishes at the source, and a viewer can check in one click.

Proof. Not detection.

https://t.co/H7RNyajWYk

[4dauth @4DAuth]

YouTube's expanded deepfake detection tool is a good thing for creators. Flag unauthorized uses of your likeness. Request takedowns. Worth having.
None of it does anything for the viewer.
When a video goes live, the viewer just has the video. No scanner, no flag queue. They either trust the source or they don't.
4DAuth is for a different scenario: the creator who wants their audience to verify them the second the content goes live.
Proof. Not detection.
https://t.co/H7RNyajWYk

[4dauth @4DAuth]

Voice-clone scams now target anyone with a public video archive.
A cosmetologist with 90K YouTube subscribers was deepfaked into a life-insurance ad she never endorsed. A Texas lawman's old YouTube clips were stitched into a viral supplement ad with a clone of his real voice. McAfee this week: one in ten Americans say they have been hit.
Three to five seconds of audio is all a cloner needs. Every video you have ever published is training data for your own impersonation.
Detection asks strangers to spot your fake. Proof lets you sign what's actually yours.
Proof. Not detection.
https://t.co/H7RNyajWYk

[4dauth @4DAuth]

Viewers can't verify content on a platform by trusting the platform to police itself. Grok is still producing nonconsensual deepfakes on X, four months after a pledge to stop. Apple threatened to pull it from the App Store. French authorities are calling X executives to Paris this week.

Verification has to come from somewhere the platform doesn't control. It has to come from the creator. That's what 4DAuth is built for. Every video a creator publishes includes a signed card. The card belongs to the creator, not the platform. Checking it confirms the video is the one they made.

Proof. Not detection.

[4dauth @4DAuth]

Viewers have a harder time than ever telling real video from AI-generated video. Microsoft's new Safer Internet Day report: teens correctly identify AI deepfakes 25% of the time, down from 46% a year ago.
4DAuth is a different kind of verification, built for viewers. Every video a creator publishes includes a signed card. Checking the card confirms the video is the one the creator made.
Proof. Not detection.

[4dauth @4DAuth]

https://t.co/AE6zk6Km3G

[4dauth @4DAuth]

A creator with 90,000 subscribers had her face deepfaked into a YouTube ad selling life insurance she never endorsed. Her image, someone else's voice. Her response: "What do we do to protect ourselves when stuff like this happens?"
That question is why 4DAuth exists. Creators shouldn't have to hope platforms catch fakes. They should be able to hand their audience a way to verify any video themselves.
We're building that.
https://t.co/H7RNyajWYk