The Terrifying Evolution of AI Video: Lessons from INMU KING and Viral Face-Swaps

A 15-second video can be made for a few hundred yen and go viral on social media.

Moreover, it works just by saying "replace this face" on an existing video. It's practically like Anpanman.

https://x.com/ChibaNyan_Music/status/2090748915272622439

The trigger was Chiba Nyan's "Omae Usokusendayo Sugi" appearing on my timeline. I laughed normally. But while laughing, I was distracted by something else.

GIF

The face doesn't break until the very end.

Today, I will organize what is currently happening in the AI video world. It's a bit long, so saving it for later is recommended.

The Strongest AI Video Tools Now Are Made by TikTok's Parent Company

In this field, ByteDance's (the TikTok company) "Seedance" is currently considered quite powerful.

A Brief History

• June 2025: First generation appears
• February 2026: 2.0 released. This is where it truly transformed.
• June 2026: 2.5 announced. Available from late July.

What it can do

If you provide a face photo, it can move while maintaining that person's face. Furthermore, in version 2.5, you can upload up to 30 reference images and 10 videos at once.

The movement, camera work, and effects remain the same. Only the person is swapped. And yet, a 5-second video costs about a few dozen yen.

What used to be called "deepfakes" and required specialized software has now become just one feature of standard video generation AI.

However

Others like Google's Veo, China's Kling, and Alibaba's WAN are running in the same space, and rankings change frequently. There is no "absolute king" yet.

What Tool Was That MV Made With? (Speculation)

To be clear, it hasn't been disclosed which tool was used for that MV. So I won't state it as fact.

That said, my view is that "Seedance is highly likely."

There are two reasons for this.

First, as mentioned, "Seedance is incredible" right now. Second, I actually heard it.

Just the day before yesterday, when I met someone from an AI startup, they said, "The updated Seedance is too crazy." After that, when I tried it myself, it was indeed crazy. The timing fits perfectly.

However, I cannot be certain.

(*According to my research, the person involved in producing Chiba Nyan's MV seems to handle other AIs as well. So please read this as speculation.)

From a Management Perspective, What is Breaking?

In video production, MVs were one of the most expensive areas. Locations, casting, equipment, editing. Hundreds of thousands to millions of yen are normal.

Now, an era has arrived where you just plug in a photo and a prompt and pay a few hundred yen.

And as discussed recently, voice AI is also showing otherworldly growth.

https://www.oricon.co.jp/news/2456912/full/#google_vignette

It has become possible to make famous voice actors say any lines you want, leading to a flood of content on TikTok and prompting the Japan Actors Union to issue a statement.

The Defensive Side is Moving at the Same Speed

As production becomes faster, mechanisms for detection and rights management are also advancing rapidly.

1. The Government is Drawing Lines

Just recently, the Ministry of Justice published guidelines on how current laws apply to generative AI and publicity rights. Notably, it explicitly stated that "voice" is also subject to protection, not just faces.

2. YouTube Created a "Protect Your Face" Feature

A system to find videos where your face is used without permission via AI and request deletion.

• Started in October 2025 for creators monetizing on YouTube.
• Registration requires an ID photo and a short video of your face.
• Approval takes up to 5 days.

By the way, I just registered for this last week. I received a notice from my agency, UUUM, and finished the procedure.

Note that detection doesn't guarantee deletion, and YouTube admits that non-AI videos might be flagged. It's not perfect. Still, we have entered an era where those being "used without permission" must also defend themselves.

Summary: Three Things Happening Simultaneously

1. Production costs are heading toward zero. Even MVs are now in the hands of individuals.
2. Faces and voices are being solidified as clear subjects of rights. It's not a gray area; if the other party acts, it will be deleted and liability will occur.
3. These two are moving at the same speed. As the weapons for creation get stronger, the weapons for defense are also being distributed.

That said, it's not a matter of "AI is dangerous, don't use it" or "anything goes." However, since AI evolution is only accelerating, business owners should keep up with the latest AI information.

...Well, I've written a lot of serious stuff.

Regarding the original video "INMU KING" posted on August 10th.

GIF

I'll refrain from introducing the content, but...

Making an MV of this quality with AI is quite something!

(I won't point out the MV details. The reason is that if I point it out, it's obvious, and even if I don't, it's obvious.)

I usually work as an SNS management and video consultant for companies at OCTOPUS Inc., and recently I've been supporting the introduction of AI-based article and video production. If you don't know how much your company can do, feel free to contact me via X DM or info@kattu0403.com.

*This article is based on information as of August 2026. AI versions and prices fluctuate wildly, so please be careful. The production tools for specific MVs are not disclosed and are the author's speculation.

References: Ministry of Justice "Interpretative Guidelines on Infringement of Publicity Rights by Generative AI" (August 7, 2026) / YouTube Help "Similarity Detection" / ITmedia NEWS (October 22, 2025) / TechNode (July 31, 2026) etc.