The Flood Made Everything Free. So Now We Pay for Proof.
Watch · Listen · Read
Listen to this article, read by TAPE9.
I look at the intersection of business, technology, and messaging regularly through three lenses. Through the first, I watch how organizations run their operations and reward the work that feeds them. Through the second, I track where technology and markets are actually shifting, not where the press release says they are. Through the third, I listen to the language, because the words an industry picks tell you what it wants you to believe before the data catches up.
This week all three point at the same quiet event. On July 15, Tidal stops paying royalties on any track it decides was made entirely by a machine.
That sounds like a music story. It is not. It is the leading edge of a pricing change that is showing up in publishing, film, science, software, and hiring at the same moment. When the cost of generating something falls to nearly zero, the thing itself stops being worth paying for, and the proof that a human made it becomes the thing you pay for instead.
I should admit a bias here. I called this a while ago. In 2017 I wrote that as the virtual becomes a commodity, true reality would come at a premium, that we would eventually pay real money for the physical world we took for granted and left in the digital dust.1 My example was the last standing antique wooden roller coaster, the kind of thing people would travel to ride in person, for real. I did not expect the first place that prediction came true to be a royalty rule. But here we are, and this time there are receipts.
What are these platforms actually re-pricing?
They have stopped paying for the song. They have started paying for proof that a person made it.
Tidal's new policy does not ban AI music. It lets the tracks stay, tags them with an AI label, and cuts them off from royalties and direct-to-fan sales beginning July 15, 2026.2 The company is explicit that it wants royalties to go to work produced, written, and performed by people. Read that as an operations decision rather than a moral one. Tidal is rewiring who its payout mechanism rewards.
The reason is a volume that no reward pool was built to survive. Deezer, which runs its own AI detection, reports that it now takes in roughly 75,000 fully AI-generated tracks a day, about 44 percent of everything uploaded, up from around 10,000 a day at the start of 2025.3 Deezer sells that detection technology, so treat the exact figure as its own measurement rather than an independent audit. The direction of travel is corroborated everywhere else.
Here is the part that reframes the whole debate. On Deezer, AI-generated music is 44 percent of uploads but only 1 to 3 percent of actual listening, and roughly 85 percent of those streams are flagged as fraudulent.
We now have that in a courtroom. In March 2026, Michael Smith of North Carolina pleaded guilty in what prosecutors called the first United States criminal case for AI-assisted streaming fraud.4 He created hundreds of thousands of AI songs and used roughly a thousand bot accounts to stream them billions of times, collecting more than eight million dollars in royalties before the Mechanical Licensing Collective and the FBI unwound it. He is scheduled for sentencing on July 29. That is the "farm a few pennies from each track and repeat" model, run at industrial scale, ending in a federal court.
The same move is already visible in security, which is where this stops being someone else's problem. When the curl project was buried under AI-generated vulnerability reports that read as technical but found nothing real, its maintainer did not build a better filter first. He shut down the bug bounty.5 Remove the payout, remove the incentive to submit machine-generated noise. It is the identical logic to Tidal's. And it carries a warning the music platforms should read closely: curl reopened the program a month later, not because the flood stopped, but because the AI reports had become good enough to be worth reading. Cutting the money removed the money. It did not remove the volume.
That lever only exists when one party controls the payout. Watch what happens where no one does. Recruiters now see roughly 11,000 job applications submitted to LinkedIn every minute, up about 45 percent in a year, with generative AI feeding the surge.6 There is no single pool to switch off, so there is no demonetization move to make. Employers reach for detection instead and end up, in one recruiter's phrase, in an AI-versus-AI standoff. Where a gatekeeper owns the payout, it can re-price around human origin. Where none does, the flood simply degrades the mechanism until no one trusts it.
If the machine makes it for free, what is left to sell?
A new product category is forming across every creative and technical field, and the product is provenance itself.
Start with the engine. Suno, the most-used AI music generator, reported two million paying subscribers and 300 million dollars in annual recurring revenue by early 2026, says more than 100 million people have tried it, and, by its own investor deck, generates around seven million songs a day, the equivalent of Spotify's entire catalog every two weeks.7 Gregoire Gensollen, a film and television producer whose series I discussed with him on the Music Evolves podcast, cited nearly identical figures. The cost of producing a finished, sung, mixed track has effectively gone to zero.
Gensollen ran the experiment most of the industry is only talking about. One episode of his show sets five human strangers, given six hours to write a song together, against a host given thirty minutes to make one with AI. What struck him was not the output but the friction. The humans argued over which beat, which melody, whose experience the song would carry, and that struggle was the thing worth filming. His line stayed with me: he said he would not remember which tool made the track, but he would remember the human moments. AI, as he put it, produces an average of what already exists. The top of any field is still human.
So the market is building a way to certify exactly that. The Authors Guild now sells a "Human Authored" certification, opened to all United States authors in March 2026, ten dollars a title, with identity verification and a public database anyone can check.8 That is the mirror image of Tidal. One side strips the money out of the machine. The other side certifies, and charges for, the human. YouTube made the same move a full year before Tidal, renaming its rules to demonetize "inauthentic" mass-produced content in July 2025 and later terminating channels built on AI-generated fake movie trailers.9 Amazon capped daily self-publishing uploads and required AI disclosure after AI-written guides put dangerously wrong information in front of readers.10 Even Suno is bolting provenance onto itself, adding identity-verified voice cloning, while Warner's settlement with the company gives artists control over their name, voice, and likeness. Authentication is becoming infrastructure.
In music and video, the flood is a nuisance you scroll past. In software and science, the same economics produce something closer to a weapon. Same mechanism, very different blast radius.
Who taught us to want “authentic”?
The word doing the most work in this whole story is not "AI." It is "human."
Consider how fast the language moved. Merriam-Webster named "slop" its 2025 word of the year, defining it as low-quality digital content produced in quantity by AI.13 Its president explained the pick by saying people want things that are real, things that are genuine. A dictionary does not choose a word like that unless the culture has already decided how it feels.
Watch the relabeling underneath it. YouTube did not announce a new AI policy so much as rename its "repetitious content" rule to "inauthentic content," which quietly shifts the question from "is this a duplicate" to "did a person make this." The industry coined "slopsquatting" and "workslop" in the same stretch. Language is where the governance actually happens, because how you name the flood decides what you are permitted to do about it.
Now notice who is reaching for the same word. Tidal, Deezer, Qobuz, the Authors Guild, and Amazon are all selling versions of one descriptor: human-made, authentic, original. It is becoming a premium tag. Meanwhile NewsGuard tracks thousands of AI content-farm sites, most built purely to catch advertising money,14 and the published story of an AI creative revolution sits awkwardly beside the data we can verify, which shows extraction, fraud, and listening rates in the low single digits. The narrative we are handed and the numbers we can check are not the same narrative.
The people who work as filters see the gap most clearly. DJ Sam Young, who runs a label and sorts through demo submissions every week, told me on the Music Evolves podcast that he can hear the sameness almost instantly, and asked the question the industry keeps dancing around: why do we need fifty versions of the same thing? He is not against the tools. He wants technology to free people to create and objects only when it is used to replace the creative act, which is close to the exact line Tidal claims to be drawing. He also noted that the download platforms serving DJs have started labeling AI-assisted tracks, so the labeling reflex has already reached the creator tier, not just the streaming giants.
That leaves the questions that actually matter. If "human" is now a premium label, who gets to certify it, and what is that certification actually checking? The Authors Guild mark, to its credit, verifies a person's identity, not the absence of AI in the work, which are two very different promises. Is "authentic" turning into a value, the thing people genuinely seek out, or into a price tag, a badge you buy? And who is setting that price, the audience or the platform?
Reality came at a premium. Who's collecting it?
The three lenses meet at a single move. The platforms are re-pricing their payouts around human origin. The market is building products that certify it. The language is teaching us to want it. Put those together and you do not get a defense of artists. You get a paywall around authenticity, sold to us as virtue.
I want to be careful here, because I predicted this destination and I still believe in it. A while back I wrote that once the artificial became a commodity, the real would command a premium, and that we would pay to touch the physical world we had abandoned. That has happened. What I did not say, because in 2017 I assumed it, was that the premium would be something people chose. The wooden roller coaster was valuable because someone wanted to ride it. That was demand.
What is arriving now is not demand. Remember the number that should stop everyone cold: on the one platform measuring it openly, machine-made music is a plurality of everything uploaded and a rounding error of everything actually played. The audience has not asked for the human premium. It is being installed above them, by the businesses that collect the fee. That is the tell. When a market prices a scarcity its own customers are not yet demanding, the scarcity is being manufactured, not discovered.
And it is not years away. It is this month. Tidal flips the switch on July 15. The Authors Guild is already selling a human badge for ten dollars a title. The certification layer is being poured right now, while the word "authentic" is still fresh enough to feel like a value instead of a line item.
So follow the money, because that is the real question. Someone owns the detector that rules your track too artificial to earn. Someone sells the badge that says your book was written by a person. Someone runs the identity check that turns "I am real" into a paid tier. The people who actually do the filtering, the DJ who hears the fiftieth copy of the same song, the producer who remembers the human moment and forgets the tool, are not the ones positioned to monetize it. The house is.
I still think reality will command a premium. I am no longer sure we, the creators, are the ones who will be paid for it. So when proof of human becomes a product, ask the only question that resolves the rest: who is making the money, and who handed them the right to decide what counts as real?
Sean Martin, CISSP, is a cybersecurity market analyst, content strategist, and go-to-market advisor with more than 30 years of experience across engineering, product development, marketing, and media. He is co-founder of ITSPmagazine and Studio C60, host of the Redefining CyberSecurity Podcast and Music Evolves Podcast, and writes Lens Four at seanmartin.com.
Subscribe to Lens Four · Where business, innovation, and messaging come into focus.
Topics Covered in This Analysis
AI-generated music, Tidal AI policy, AI royalty demonetization, Deezer AI detection, AI music flooding, streaming fraud, Michael Smith streaming fraud case, Mechanical Licensing Collective, Suno, AI music generation, provenance economy, proof of human origin, Human Authored certification, Authors Guild, C2PA content provenance, content authentication, YouTube inauthentic content policy, AI slop, mass-produced content demonetization, Amazon KDP AI policy, AI-generated books, curl bug bounty, AI vulnerability reports, slopsquatting, AI package hallucination, software supply chain security, AI paper mills, fraudulent research, cancer research integrity, AI job applications, recruiting flood, LinkedIn hiring, detection arms race, Merriam-Webster word of the year slop, NewsGuard AI content farms, authenticity premium, human-made content, DJ Sam Young, Gregoire Gensollen, Music Evolves podcast, Sean Martin, Lens Four.
References
- Sean Martin, "When Virtual Reality Is A Commodity, Will True Reality Come At A Premium?" (2017). medium.com/@sean-martin
- Tidal, AI Policy (official support page). support.tidal.com
- Deezer Newsroom, "AI-generated tracks now represent 44% of all new uploaded music" (April 20, 2026). newsroom-deezer.com
- U.S. Department of Justice, SDNY, "North Carolina Man Pleads Guilty To Music Streaming Fraud Aided By Artificial Intelligence" (March 19, 2026). justice.gov
- BleepingComputer, "Curl ending bug bounty program after flood of AI slop reports." bleepingcomputer.com Reopening and quality shift: cybernews.com
- eWeek, "Job Seekers, Some Using AI, Flood LinkedIn With 11,000 Applications a Minute." eweek.com
- TechCrunch, "AI music generator Suno hits 2M paid subscribers and $300M in ARR" (Feb 27, 2026). techcrunch.com Users and daily volume: forbes.com
- The Authors Guild, "Human Authored" certification. authorsguild.org
- TechCrunch, "YouTube prepares crackdown on 'mass-produced' and 'repetitive' videos" (July 2025). techcrunch.com
- The Authors Guild, "Amazon Adds to KDP Generative AI Policy, Caps Daily Self-Publishing Uploads." authorsguild.org
- Aikido Security, "Slopsquatting: the AI package hallucination attack" (cites USENIX Security 2025). aikido.dev
- KFF, "How AI Can Both Detect and Enable Fraudulent Research." kff.org
- PBS NewsHour, "Merriam-Webster's word of the year for 2025 is AI 'slop'." pbs.org
- NewsGuard, AI Tracking Center. newsguardtech.com
Frequently Asked Questions
What is Tidal's new AI music policy?
Beginning July 15, 2026, Tidal stops attributing royalties to tracks it identifies as fully AI-generated, tags them with an AI label, and blocks them from direct-to-fan sales. The tracks are not removed. The policy re-prices the payout so that royalties flow to work made by people.
How much streaming music is AI-generated?
By Deezer's own detection, roughly 44 percent of tracks uploaded to its platform each day are fully AI-generated, about 75,000 a day. That same AI music accounts for only 1 to 3 percent of actual listening, and around 85 percent of those streams are flagged as fraudulent, which suggests the flood is driven by royalty farming rather than audience demand.
Is this "AI flood" only a music problem?
No. The same pattern appears in publishing (Amazon KDP caps, the Authors Guild "Human Authored" mark), film and video (YouTube demonetizing inauthentic content), software (slopsquatting and the curl bug-bounty shutdown), science (AI paper mills), and hiring (AI-generated job applications). When generation cost approaches zero, proof of human origin becomes the thing being priced.
What is "slopsquatting"?
Slopsquatting is a supply-chain attack that exploits AI coding assistants. Research found that roughly 20 percent of the packages such assistants recommend do not exist. Attackers register those hallucinated package names, so that developers who trust the AI's suggestion download malicious code. The term was coined by a Python Software Foundation security developer.
What does the "Human Authored" certification actually verify?
The Authors Guild's "Human Authored" mark verifies the author's identity and their attestation that the work is human-written, and lists the title in a public database. It confirms who is behind the book, not the technical absence of AI in the text. Those are two different promises.
