Sony Music vs AI Deepfakes: 260,000 Unauthorized Tracks Target Its Artists

Artificial intelligence has moved from being a futuristic music-production tool to one of the biggest business and rights challenges facing the modern music industry. Now, the scale of the problem is becoming clearer.

Sony Music Entertainment has asked digital platforms to remove more than 260,000 AI-generated tracks impersonating artists on its roster as of the end of September 2026. That figure is almost double the more than 135,000 removal requests Sony had made by the end of March. The latest figures were reported by the Financial Times and subsequently confirmed by multiple music-industry publications.

Artists whose voices or identities have reportedly been targeted include Harry Styles, Adele, Britney Spears, Queen and Michael Jackson.

Sony says unauthorized generative-AI deepfake tracks are happening at scale, harming artists, misleading fans and undermining the ethical use of AI. The company has promised to continue enforcing its artists' rights while pushing streaming platforms to respond more quickly and prevent removed material from returning.

But the 260,000 figure is about more than one major label fighting fake songs. It highlights a rapidly expanding problem involving AI voice cloning, artist impersonation, streaming fraud, distribution platforms, copyright, personality rights, and royalty manipulation.

Sony's AI Deepfake Takedowns Have Nearly Doubled

The most striking part of Sony's latest disclosure is the speed at which the numbers have increased.

By the end of March 2026, Sony had requested the removal of more than 135,000 AI-generated tracks. Six months later, that number had passed 260,000.

That means Sony requested the removal of more than 125,000 additional tracks between April and September alone.

Importantly, these numbers represent takedown requests, not necessarily confirmed removals.

Sony has not disclosed how many of the 260,000 tracks were ultimately removed, how quickly platforms responded or exactly which streaming and digital services received the requests. Therefore, it would be inaccurate to say that 260,000 AI deepfake songs have definitely been removed from streaming services.

Instead, the figure represents the scale of the enforcement workload Sony says it is facing.

The company told the Financial Times that it will continue enforcing its artists' rights and expects platforms to act quickly while preventing the same material from reappearing. At the same time, Sony acknowledged that the battle is becoming increasingly difficult.

Harry Styles, Adele, Britney Spears and Michael Jackson Among Targets

AI deepfakes become particularly problematic when they use the identity of an established artist.

Sony says its enforcement efforts have included unauthorized tracks imitating artists such as Harry Styles, Adele, Britney Spears, Queen and Michael Jackson.

The technology can reproduce vocal characteristics closely enough to create recordings that may initially sound convincing to listeners.

That creates an obvious problem.

A fan searching for a new Harry Styles, Adele or Michael Jackson song on a streaming platform could encounter an AI-generated recording that appears to be associated with the artist. Unless the content is clearly labeled, the listener may not immediately understand that the performer never recorded the song.

The issue is particularly sensitive with deceased artists such as Michael Jackson, because AI can potentially create entirely new performances that the artist never recorded or approved.

For living artists, meanwhile, unauthorized voice cloning can effectively turn a recognizable artistic identity into something that third parties can commercially exploit.

AI Voice Cloning Is Not the Same as Ordinary AI Music

It is important to distinguish between different forms of artificial intelligence in music.

An artist using AI as part of a songwriting, production or mixing process is not necessarily creating a deepfake.

Likewise, a producer using AI-assisted software to generate backing vocals or manipulate an existing performance is very different from uploading a song designed to make listeners believe that a famous artist performed it.

The problem Sony is describing involves unauthorized imitation of an artist's voice or likeness.

That is why the debate increasingly involves more than traditional copyright.

An AI-generated song that sounds like a particular singer can raise questions about the artist's name, voice, identity, publicity rights and commercial reputation. Copyright may be relevant to the underlying recording or composition, but an unauthorized vocal imitation can create additional legal questions.

Spotify, for example, distinguishes between unauthorized AI voice clones and another type of impersonation problem: music that is fraudulently delivered to the profile of another artist.

Those are related problems, but they are not identical.

The Streaming Platform Problem

One reason AI deepfakes are difficult to control is the enormous scale of digital distribution.

A traditional counterfeit recording might require significant resources to manufacture and distribute.

AI changes that equation.

Someone can generate music, create an imitation vocal, distribute multiple versions through digital channels and potentially upload huge quantities of material at a speed that traditional enforcement systems were never designed to handle.

Even after a track is removed, another version can potentially appear under a different title, account or distributor.

This is why Sony is specifically asking platforms not only to remove unauthorized tracks but also to prevent the same content from reappearing.

The industry therefore faces a continuing cycle:

Generate → distribute → detect → remove → re-upload.

If generation remains faster than detection and enforcement, rights holders can end up playing an endless game of catch-up.

AI Music and Streaming Fraud Are Becoming Connected

The deepfake problem is also part of a much larger streaming-fraud issue.

Not every AI-generated song is a deepfake, and not every AI-generated song is fraudulent. However, the enormous volume of synthetic music being delivered to streaming platforms creates opportunities for people attempting to manipulate royalty systems.

Deezer has provided some of the clearest evidence of the scale.

In June 2026, the platform received an average of around 90,000 fully AI-generated tracks per day, with AI music accounting for more than half of new daily deliveries at peak levels.

Yet the percentage of listening going to fully AI-generated music remained much smaller.

Deezer said AI-generated tracks accounted for only around 1% to 3% of total streams, while up to 85% of streams on fully AI-generated tracks were identified as fraudulent in 2025. Those fraudulent streams were excluded from royalty payments.

That distinction matters.

The problem isn't simply that AI music is becoming popular.

The bigger concern is that some people can generate huge volumes of inexpensive music and then attempt to manipulate streaming systems to generate artificial royalties.

Why Streaming Fraud Hurts Human Artists

Streaming royalties are distributed from a finite pool of revenue.

When fraudulent streams are counted as legitimate consumption, money can be redirected toward recordings that have not actually attracted genuine listeners.

That can reduce the effective value of legitimate streams.

AI can make this problem more scalable because generating thousands of tracks no longer requires thousands of recording sessions, musicians or traditional production budgets.

A bad actor could theoretically create large amounts of synthetic music, upload it through distribution services and use artificial streaming activity to generate revenue.

Sony executives have previously warned about the scale of streaming fraud. The Financial Times reported that Sony's global digital business president Dennis Kooker estimated fraudulent streams could represent a significant portion of overall streaming activity, while industry estimates have placed the potential annual cost of streaming fraud in the billions.

That is why AI deepfakes and streaming manipulation increasingly appear in the same industry conversation.

Sony Is Not Fighting AI Itself

It would be misleading to characterize Sony's position as simply being “against AI.”

The music industry is increasingly separating authorized AI use from unauthorized exploitation.

Sony Music joined the Alliance for Responsible Innovation in the Arts & Media in September, becoming the first music company to join the coalition. Sony said generative AI presents both significant opportunities and challenges and that it wants to help shape policies that protect creativity and consumers while supporting innovation.

Sony is also part of a much broader industry movement toward rules governing AI-generated recordings.

In July, Sony, Universal Music Group, Warner Music Group, Believe, BMG and other companies proposed global principles for determining when AI-assisted recordings should qualify for official music charts. Their proposal says AI services should be properly authorized, recordings should remain substantially human-made and tracks should not involve stream or chart manipulation.

The emerging message is therefore less “AI must disappear” and more:

AI should be authorized, transparent, properly labeled and respectful of creators' rights.

The Industry Is Building New AI Detection Systems

The growing number of AI tracks is forcing streaming companies and distributors to improve detection technology.

Deezer has been particularly aggressive in this area. Its AI detection system can identify fully AI-generated recordings from major generative music platforms, including Suno and Udio. The company has also made its detection technology commercially available to other organizations in the music ecosystem.

Deezer says it has detected and tagged more than 13.4 million AI-generated tracks in 2025 and automatically removes detected AI-generated songs from algorithmic recommendations and editorial playlists.

This suggests that the future of streaming may involve an increasingly sophisticated layer of automated content classification.

Instead of simply asking whether a song is available, platforms may increasingly need to know:

• Who created it?
• Was AI used?
• Which AI tools were involved?
• Was an artist's voice cloned?
• Was the artist authorized?
• Is the track being artificially streamed?
• Does it meet the platform's distribution requirements?
• Should it qualify for charts?

The Suno and Udio Battles Add Another Dimension

The Sony deepfake problem is also developing alongside major legal battles involving AI music companies.

Sony and Universal Music Group have sued Suno, arguing that its AI models were developed using copyrighted recordings without authorization.

The latest litigation has also focused on Suno's v6 models and how information from previous AI systems may have contributed to their development.

Suno disputes the allegations and says v6 was trained using licensed partner content, community interactions including user creations and preference signals, and the accumulated technical knowledge of its team.

Suno has also acknowledged in court filings that audio data was obtained from YouTube using YT-DLP for training purposes, while challenging aspects of the labels' legal claims.

This is a different issue from unauthorized artist deepfakes, but the cases demonstrate how many separate legal questions AI is creating for the music business.

What Does This Mean for Artists?

For musicians, the rise of AI deepfakes creates both opportunities and risks.

Artists can potentially use AI to speed up production, experiment with sounds and develop new creative workflows.

But their voices can also become valuable digital assets that others attempt to exploit without permission.

That could lead to new types of licensing agreements in which artists explicitly authorize the use of their voices for AI projects.

Instead of an anonymous third party cloning a singer's voice, an artist might voluntarily license an authorized digital replica and receive compensation whenever it is used.

That model could turn voice cloning from a threat into a new revenue stream — but only if artists retain meaningful control.

Why the 260,000 Figure Matters

Sony's latest number is significant because it provides a rare glimpse into the sheer scale of unauthorized AI music enforcement.

The company requested the removal of more than 75,000 AI deepfake tracks by March 2025, then surpassed 135,000 by March 2026 and more than 260,000 by September 2026.

The growth curve is difficult to ignore.

More importantly, the 260,000 figure does not represent every AI-generated recording in existence. It represents tracks Sony identified as requiring platform intervention because they allegedly impersonated its artists.

That means the wider volume of synthetic music is considerably larger.

The challenge for the music business is now moving from whether AI music exists to how the industry can identify, authenticate, monetize, and regulate it at internet scale.

The Future of AI Music Will Depend on Trust

Streaming has always depended on trust.

Fans trust that an artist profile actually belongs to the artist. Labels trust distributors to deliver legitimate recordings. Platforms trust rights holders to provide accurate metadata. Artists trust the royalty system to reward genuine consumption.

AI threatens to weaken those assumptions.

If a fan cannot tell whether a new song is authentic, if an artist cannot control their own voice, and if platforms cannot distinguish genuine streams from automated activity, the value of the entire ecosystem becomes harder to protect.

That is why Sony's fight against AI deepfakes is bigger than a dispute involving 260,000 tracks.

It is ultimately a fight over identity, authenticity, attribution, and money.

The technology itself is not going away. In fact, the volume of AI-generated music is increasing rapidly.

The music industry's next challenge will be building systems capable of allowing legitimate AI creativity to grow while making unauthorized imitation, fraud and deception much harder to profit from.

For Sony Music, that battle has already reached an enormous scale — and the company's warning that the fight is becoming an “uphill struggle” suggests the industry may only be seeing the beginning of the AI deepfake problem.