16:08
Hi. Welcome to another episode of ColdFusion. Over the past year, it's become very clear that AI is infiltrating the music industry. AI is coming for your gigs. It's already writing songs, performing music, and getting better and better by the day. Our CNBC senior media tech correspondent, Julia Boorstin, has been digging into how AI is disrupting a range of different industries. AI music prompters have signed huge deals, and AI R&B artist, Zenaya Monae, signed a multi-billion dollar record deal with Howard Media on behalf of her prompter, Nikki Jones.
Other AI groups, like The Velvet Sundown, have also gone viral. So, of course, there's countless examples of AI music flooding platforms with 60,000 songs a day detected on the streaming platform Deezer, and almost 34% of tracks uploaded to Spotify daily are AI. But, what about something a little different? A wacky story about how an AI band turned into a real band with real humans, and is now playing live shows. This is the first case I've ever heard of where an AI band has turned real. In this episode, we explore the story, plus check out all the wild things that have happened over the past few months in AI and robotics. From a grandma wrongly imprisoned because of
AI, Chinese robots running a marathon and breaking the human record, AI video generation taking a massive leap forward, so much so that people believe that animation is dead, an out-of-control restaurant robot that couldn't stop dancing, cleaning robots going mainstream in China, plus much, much, much more. This is an absolutely packed episode, so let's get into it. You are watching ColdFusion TV. Our story begins in mid-2025. A mysterious kawaii Japanese metal band by the name of Neon Oni showed up on streaming platforms.
They had over 80,000 monthly listeners on Spotify with their top song having over 1.2 million streams. The band had seven members, two female singers, two guitarists, a bass player, a drummer, and a producer. Plenty of fans had the band in their Spotify wrapped top five, but in order to really sell the group to people, you can't just have the music. So, in September of 2025, the first music videos of the band appeared on YouTube with other snippets appearing on Instagram. At first, fans liked what they saw. One comment reading, quote, "I'm here before this band gets famous. Absolutely amazing." End quote. The Japanese band seemed like an upcoming force in the indie metal scene. Fans connected with them and their
merch began to sell. But the praise wouldn't last. As more content of the band began to be uploaded, fans began to notice strange artifacts typical of AI-generated content. Among the community, a great debate raged. Facebook groups and Reddit threads began discussing what on Earth was going on. They were thinking to themselves, was this band that we connected with AI all along? Have we all been fooled? As the AI accusations began to spread, the band started to delete the comments. The jig was up and fans felt like they had been scammed.
I think they're AI, bro. They have to be. Someone said if it's AI, it's the best AI on the internet and I want more. I think it's AI. It's really good AI. people are thinking it's AI. That's really scary. As investigations continued, some online sleuths determined that the so-called Japanese band was allegedly actually an AI music prompter from Europe using Suno. So, one guy had fooled almost 100,000 people. That could have been the end of the story, but it wasn't. The maker of the group broke their silence and publicly stated, quote, "What began as a small experiment born inside a machine was never meant to be heard, let alone grow into something real. But, people listened, people cared, and again they asked
for more. Originally created in the digital world, it grew faster than expected. New music, visuals, merchandise, and stories all shaped by a community that wanted something more than a virtual band. Eventually, the biggest request became impossible to ignore. "Make it real." End quote. So, as you've probably gathered, in a strange twist, the band would become real. Seven musicians from actual Tokyo bands were recruited to perform the AI-generated songs live. They've now played several live shows and continue to do so till this day. This is the first high-profile case of music going from AI to human, but it's not going to be the last. The AI prompter of the digital singer, Zenaya
Monae, is preparing to hire a singer and turn the AI-generated music into live shows. This is in spite of the backlash. But, Nikki, they thought Zenaya was a real person. Zenaya is an extension of me, so I look at her as a real person. But, you can't sing. So, what does this signify? If you don't think about the situation too much, you could say that the situation with the Japanese band Neon Honey is the same thing as any other band performing a cover of someone else's music live. And the Zenaya Monae situation is like an artist performing a song live that was written by a ghostwriter.
Both of those things have been happening for decades. There's an argument to be made that if the audience likes it, who really cares? But, in a way, it's bypassing the purpose of what music is supposed to be. Its primary goal is not to create money via popular demand, but it's the expression of human life experience. Music is born from frustration, joy, anger, whatever emotion that may be, and it resonates with people that are feeling the same way. Generative AI makes music a commodity that merely mimics those things. Some people care, others don't. I've been a guitarist for over 20 years. I may have my own bias here. But, regardless, I'm not going to argue which side you should be on. You are
your own person and you're free to believe whatever you like. But for those who don't want a future where all music is AI, it's not like there's nothing we can do. The music streaming platform Deezer already has a tool that can detect AI music with 98% accuracy. So these things exist. Platforms just need the incentive to implement them. At the end of the day, more AI content on streaming platforms means more revenue for these companies despite humans being outpaced by AI. Fortunately, Spotify has removed over 75 million AI tracks that they say were low effort. So that's a good start. With that being said, the problem goes far beyond just flooding streaming
platforms. Some generative AI music can really hurt people. Singer-songwriter Murphy Campbell had her music stolen by AI and the prompter started making money from rip-off versions of her songs. And then that AI prompter turned around and copyright claimed Murphy's own music. So now she makes nothing. An entity called Timeless Sounds IR uploaded AI generated versions of my songs to all major streaming platforms. And to do that, they fed YouTube videos of me to an AI engine that then mimicked my voice and playing. They used a distributor, which I just discovered, and that distributor's name is Vydia. And Vydia has since decided that it was going to make copyright claims on all of the videos that were used to feed the AI
engine to sound like me. I am no longer making money on YouTube. Vydia is making money on YouTube off of my own videos and me playing my own banjo in my own backyard with traditional folk songs, some from my own family. over AI generated music. It's the wild west and it could happen to anyone. But let's move on to the second part of the episode and talk about some wild things that have been happening in AI over the past few weeks. These will just be a few quick-fire stories to let you know what's happening around the world. First, we have Chinese robots that competed in a half marathon on April 19th. Over 20 teams joined.
This robot, called Lightning, an entry from the smartphone manufacturer Honor, won. It finished in 50 minutes and beat the human world record by 6 minutes. This other robot reportedly came last. And some, well, they just didn't make it. AI video generation has also moved forward again, as users get their hands around seed dance, too. Here's a couple of short Japanese ones from the user AI Vitamin C. Viewers couldn't believe that it was AI. Now, it's not only music that's being destroyed by deceptive people using AI.
Online scams, for example, have surged with the rise of AI tools, too. That's why I started using DeleteMe to protect my online presence. An entire industry is built around data brokers that gather and sell personal information. Data like your name, address, phone number, and even details about your relatives, often to anyone willing to pay. That kind of access can fuel scams, phishing attempts, and a loss of privacy. For me, it's one of those things where you don't realize how widespread the issue is until you actually see it. And then having something that works quietly in the background just makes sense. For example, when I ran a report on myself, it shows that it had cleaned my information from 47 listings and
currently has two removals in progress. And if I head to the dashboard, it shows how my privacy is being actively protected with multiple exposures found and removed. So, it's a hands-free subscription service that removes your personal information from hundreds of data broker and people search websites. And it keeps monitoring and deleting as it resurfaces over time. And what really stands out is how straightforward it is. You submit your details once and within a week you start receiving privacy reports showing where your information was found and what's been removed. It's beyond a one-off cleanup. It's ongoing protection from these sneaky guys. They've been doing this for more than 15 years.
Everything is handled in house. And if you need something specific taken down, you can reach out to real privacy advisers for support. If you want to take control of your personal data, you can get 20% off Delete Me consumer plans by going to joindeleteme.com/codefusion and use code codefusion at checkout. That's joindeleteme.com/codefusion. Link in the description or you can use the QR code on screen. Thanks to Delete Me for supporting ColdFusion. Now, back to the episode. Marko Slavnic sent X into a frenzy with his AI-generated short. 5 million views in a day.
It's the first time a viral AI-generated short has crossed the slop threshold. Some people are excited. Some were upset. But, let's see it for ourselves. It runs for about 40 seconds. What is it? It's one of those things they talk into. You mean a phone? Whatever it's called. It's called a phone and that's not a phone. I think it's a phone. And I think you're a nitwit. Your mother's a nitwit. What did you say?
Ow! Stop What you guys looking at? A phone. I'm going to kill you. Why don't you just push the button? Yeah, genius. Why don't you just push the button? Why just push the button? YOU PUSH IT. OW! YOU PUSH IT. YOU I want to go next. In order to make it, storyboarding and shot generation was done by GPT image 2 and Nano Banana, similar to the example shown on screen.
Marker used Seed Dance 2 as the video generator, and it can only generate up to 7 seconds at a time. So, he had to generate a few times and stitch the results together. I'm sort of split here. It feels wrong to say, but the short video is pretty decent, AI or not. The pacing lands, the story makes sense, and it's reminiscent of an older, simple Pixar short. So, some of that human taste does come through. But, on the other hand, tools like Seed Dance 2 does take the place of a lot of animation work. Moreover, there was an insurmountable amount stolen from real artists to make the training data for Seed Dance 2. So, I am definitely curious to know what you guys think
about this one. Moving on, South Korea gets its first robotic monk. Let's just say that a lot of the comment section wasn't happy. Back in China, because humanoid robots are becoming so common, there's now specific robot insurance for those robots and the people around them. Next, Open AI finally updated their voice model. Now, it's not going to interrupt as much and will supposedly be more intelligent. People are impressed, but Open AI is going to need a lot more to dig themselves out of their spending
hole. The ongoing lawsuit with Elon Musk is also giving them a massive headache. French robotics company Genesis AI reveals their Genie 26.5 foundational model that powers a pretty dextrous robot. The video you're seeing is real time and was trained on 200,000 hours of human hand data. It succeeds around 65% of the time for longer chain dextrous tasks. Figure Robotics made an appearance at the White House, but before that, they unveiled their robot autonomously doing household chores.
They also claim to have scaled up production to one robot per hour. Chinese robot company Xsquare is running a home robot cleaning service for 21 US dollars a visit. How good it actually is, I'm not sure, but here's the Instagram user Italy UK China to find out. Let's see how much I spend to use a robot to clean my home in Shenzhen. As soon as it enters my home, it sees the table. So, it begins using its little hands to grab every object, even the thinnest ones. It's crazy because this is not a lab testing. We are in my home, and this might be probably the very first time AI service is applied to a real home scenario. It organizes pillows, a blanket, and whatever it finds very neatly.
I see a clear division of tasks between the human cleaner and the robot, but they also work together smoothly in some tasks. In my bedroom, it spots the laundry basket. It grabs it and carries it toward the washing machine, and then it begins loading the clothes one by one into the machine. In total, I paid 149 yuan and zero tips. While watching the robot working in my home today, I asked myself this question. Will robots replace cleaners one day? Honestly, I'm not sure, but today I saw them work as a team. The cleaner handles people and complex tasks, while the robot does the basic work. AI sentiment drops to record low, with only 17% of Americans thinking it will have a positive impact.
Surveillance AI wrongly identified a grandma, landing her in jail for 6 months. Police used AI facial recognition that tied her to a bank fraud case, arrested her at gunpoint, and imprisoned her. The AI had simply misidentified her face on a surveillance camera. An AI software known as Clearview AI was used to help crack the case, or so they thought. West Fargo investigators said the technology identified a potential suspect with similar features to Angela Lips. Police wouldn't release her until her lawyer pointed out that she was 1,200 miles away at the time of the crime. It's yet another reminder that AI isn't yet good enough to be used in such consequential applications such as surveillance.
Allegedly, while in jail, this grandma couldn't make her payments on her house and car, so sadly, she lost them both. Let's hope she has a huge payday though. Elsewhere in California, a robot couldn't stop dancing and caused chaos in a Chinese restaurant. The consulting firm McKinsey got hacked through vulnerabilities opened up by its internal AI tool. Yahoo's CEO came out saying that AI-generated slop is clogging up the company's management. A lot of Open AI's data center plans are collapsing, so they're deciding to rent hardware instead of building new data centers.
Half of US data centers planned for 2026 are expected to be delayed or canceled. One big reason is a shortage of electrical equipment such as transformers, switch gear, and batteries. US does not have the manufacturing capacity, forcing it to rely on imports. Gamers are furious that Nvidia released an AI filter for games. It comes in the form of the latest DLSS series, which was traditionally used to upscale games using AI. The move to an AI filter is controversial. Some say it destroys the original artist's vision, while others see it as the way forward.
You know, when you're a trillionaire, I guess you can do that and not really care. I wasn't planning on making any sort of video about this, but really, after the major keynote at GDC, and then he was asked again like about, you know, gamers pushing back on it. And for him to say this, Well, first of all, they're completely wrong. So, that's where I mean by when I say he's just left the building. He's like not He can't read the room cuz he's not even in the same building. He's not even in the same county. And I guess when you're sitting on your Scrooge McDuck tower of gold, you can say whatever the hell you want because we're not the consumers anymore. Over in the UK, lawmakers rejected a proposal that would legally
allow AI to use copyrighted works. Globally, Pokémon Go players were shocked to learn that 10 years on, the data was used to train robots in the real world. Around the world, there's a rush to create a standardized no AI logo. There's a growing body of evidence saying that AI may make the barrier to coding lower, but over time and the more it's used, productivity actually slows down as the code becomes riddled with errors. A website popped up where AI agents could hire humans to complete real tasks. The website turned out to be vibe-coded with AI, so its entire database immediately got hacked. But at the same time over in Australia,
we have a man that used AI and collaborated with scientists to create a personalized vaccine for his dog. This person just created a custom cancer vaccine for his dying dog using ChatGPT. The crazy part, he has zero medical background. Here's the full story. Paul Cunningham's dog Rosie was diagnosed with cancer. The vet said she had only months left. He spent thousands of dollars on chemo and surgery, but nothing seemed to work. So, Paul opened ChatGPT and decided to figure it out himself. ChatGPT suggested immunotherapy and led him to a genomics lab at the University of New South Wales. There, he paid $3,000 to compare a fresh copy of Rosie's healthy DNA and her cancer DNA to pinpoint
exactly what had changed between the two. He then fed that data into AlphaFold and designed a custom vaccine built specifically from Rosie's own DNA. The lab manufactured it, gave it to Rosie, and within weeks the tumor had shrunk by half. The scientists were shocked, not just because it worked, but because a guy with zero medical background He figured it out on his own using AI. But the bigger question is, if this works for a dog, why can't it work for humans, too? Pretty cool. And with that, we come to the end of this episode.
I was just trying something a bit different, seeing what you guys think about it. So, what we've seen today is a pretty diverse set of outcomes for AI. There has been and will be a lot of change this year. So, what do you think about AI musicians turning into a real band? And feel free to share your comments on some of the other AI stories we talked about. And if you want to see another episode, you can check out my postmortem on the catastrophic NFT bubble. It's a fun one. Anyway, that's about it from me. My name is Degogo, and you've been watching ColdFusion. And I'll catch you again soon for the next episode. Cheers, guys.
Have a good one. ColdFusion. It's knee-syncing.
