Spotify is tightening its approach to artificial intelligence in music, stepping up efforts to protect artists from unauthorised voice cloning, fraudulent uploads and AI-generated content designed to manipulate streaming royalties. However, the streaming giant is not imposing a blanket ban on music created with artificial intelligence. Its stated approach is to protect human artistry while allowing musicians to use AI as a creative tool.
In its latest update, published on October 20, 2026, Spotify revealed that it had taken more than 50 million spam and low-quality tracks out of recommendations during the year. This is in addition to more than 75 million spammy tracks removed from the platform altogether over the preceding 12 months for serious or repeated violations of its rules. The figures highlight the scale of the problem facing streaming services as AI tools make it increasingly easy to generate and distribute music in large quantities.
For South African musicians, producers, independent labels and distributors, the changes raise important questions about how AI can be used in music production, how artists should disclose their use of the technology, and what happens when an AI-generated song imitates an established musician.
Spotify is targeting spam, not every AI-generated song
Spotify’s position is that artificial intelligence can be useful in music creation, but it must not become a means of deceiving listeners, exploiting artists or flooding the platform with content designed to generate artificial streams.
The company has repeatedly distinguished between musicians who use AI as part of a creative process and bad actors who exploit the technology to upload large volumes of repetitive, misleading or fraudulent material.
In its September 2025 announcement on stronger AI protections, Spotify said: “We envision a future where artists and producers are in control of how or if they incorporate AI into their creative processes.”
The distinction is important because AI can be used in several ways. A producer might use an AI tool to assist with mixing, generate a musical idea, clean up a vocal recording or develop an instrumental arrangement. Another uploader might use AI to generate hundreds of songs with little human involvement, imitate a famous singer without permission or exploit misleading metadata to attract listeners.
Spotify’s policies focus on the conduct surrounding the music, including impersonation, spam, deception and manipulation, rather than treating every use of AI as a violation.
Its October 2026 update also confirmed that the company continues to adjust its recommendation systems to reduce the visibility of spam and low-quality content. It combines several signals, including listener feedback, engagement, upload patterns, metadata and information from third-party AI detection tools, rather than relying on a single automated test.
How much music has Spotify removed?
Spotify’s latest figures need to be understood in two separate categories because removing a track from recommendations does not necessarily mean deleting it from the platform.
- More than 75 million spammy tracks were removed from Spotify entirely over the 12 months reported in September 2025, following serious or repeated violations of its rules.
- More than 50 million tracks identified as spam or low-quality content were removed from recommendations during 2026, according to the October 20 update.
- Spotify has continued removing content that seriously or repeatedly violates its rules, but has not published a new cumulative figure for all tracks deleted since the original announcement.
These figures should not be added together and described as 125 million deleted songs. The 50 million figure concerns recommendations, while the 75 million figure refers to tracks removed from the platform. The two measures describe different enforcement actions.
Spotify’s crackdown also extends beyond fully AI-generated music. The platform is targeting mass uploads, duplicate recordings, misleading search optimisation, spam-focused playlist networks and artificially short tracks designed to exploit the streaming system.
These practices can distort music discovery and potentially divert royalties from artists whose listeners are engaging with their work legitimately.
The rules artists need to know
For artists and their teams, the most important distinction is between using AI creatively and using it in ways that violate Spotify’s rules.
- Unauthorised voice cloning is prohibited. Spotify says it will remove music that impersonates another artist’s voice without permission, whether the clone was made with AI or another method. This can apply even when the uploader does not name the impersonated artist in the credits.
- AI use can be disclosed. Artists can declare AI contributions to lyrics, vocals, instrumental performances and production through participating distributors.
- Spam and manipulation can trigger enforcement. Repeated or serious violations involving mass uploads, duplicates, deceptive metadata and artificial streaming can result in content being removed.
- AI-generated artist identities may be labelled. Spotify has introduced AI Persona badges to identify profiles believed to represent AI-generated identities rather than real people.
- AI use does not automatically disqualify a musician. Spotify says it does not intend to punish artists simply for responsibly disclosing their use of AI.
There are exceptions to the impersonation rules, including certain forms of parody. However, artists should not assume that calling a track an AI experiment or fan-made version makes an unauthorised voice clone acceptable.
What artists must declare – and how to do it
Spotify launched its AI Credits feature in April 2026, allowing artists and rights holders to disclose specific AI-generated contributions to a song. The information is submitted through participating distributors and can appear in Song Credits on the mobile app.
The system is designed to reflect the fact that AI use is not always an all-or-nothing process. A song may feature a human vocalist, AI-assisted lyrics and an AI-generated instrumental, for example.
The credits can identify AI involvement in:
- Lyrics and songwriting contributions.
- Lead or backing vocals.
- Instrumental performances.
- Music production.
Where AI and human contributors both perform the same role, their contributions should be credited separately where the system allows it.
Artists should contact their music distributor to establish whether it supports Spotify’s AI Credits submission process. The feature remains in beta and is not available through every distributor. Disclosures are currently visible to listeners in mobile Song Credits and the Now Playing view.
Importantly, the absence of an AI credit does not prove that a track was made entirely by humans. The system relies on information supplied by artists, labels and distributors, and not all distribution partners support disclosure yet.
Spotify is also supporting work on an industry-wide disclosure standard developed through DDEX, an organisation that develops standards for exchanging digital music information. The goal is to make AI credits more consistent across the music supply chain, rather than requiring every streaming service to invent its own terminology.
For South African independent artists, this makes communication with distributors particularly important. Before releasing music, artists should confirm how AI contributions can be declared, ensure their metadata is accurate and retain permission records for any licensed voice models, samples or other third-party material.
The 1,000-stream rule and its impact on AI music
Spotify’s crackdown also intersects with its existing recording royalty eligibility threshold.
Since April 2024, a track has generally needed at least 1,000 streams within the preceding 12 months, alongside a minimum unique-listener requirement, to qualify for recorded music royalties. Spotify does not pay recording royalties on tracks that fail to meet the eligibility criteria.
In its October 2026 update, Spotify said tracks showing indicators of a high probability of being AI-generated are 10 times more likely to fall below the threshold than other tracks. The company presents this as one way its existing system helps limit attempts to divert royalties through mass-generated music.
This does not mean that AI-generated songs are automatically ineligible for royalties. Nor does it mean that every track with fewer than 1,000 streams has been identified as AI-generated. The threshold applies to recording royalty eligibility, while enforcement against fraud and spam involves separate assessments.
For independent musicians, the practical message is to focus on building genuine audiences, promoting releases legitimately and avoiding services that promise streams through artificial or suspicious activity.
Spotify introduces AI Persona labels
Spotify is also addressing a different issue: whether listeners can tell if the artist they are following is a real person.
In August 2026, the company announced AI Persona badges for profiles whose public identities appear to be AI-generated. The labels began rolling out in September and can appear on artist profiles, in search results and alongside tracks in playlists.
Artists can disclose that they represent an AI Persona through Spotify for Artists. Spotify also reviews profiles itself, rather than relying exclusively on artists to identify themselves.
Where Spotify applies a badge following its review, the artist is notified and given an opportunity to disclose their status or appeal the decision.
By default, AI Personas are excluded from editorial and algorithmic recommendations, although listeners may still encounter their music through other routes, including by following the profile. The policy concerns the identity presented by the artist, not whether a human musician used AI during recording or production.
The company has also introduced Verified by Spotify, which had verified more than 300,000 artists by October 2026. Spotify said those artists represented more than 99% of the artists listeners search for on its service, with more than 70% being independent musicians.
Another measure, Artist Profile Protection, allows participating artists to review releases that list them as a performer before the music appears on their profile. Spotify says it plans to make this protection available to all artists by the end of 2026.
The tools are intended to reduce confusion when music is uploaded under the wrong artist’s name or an AI-generated identity is presented as though it belongs to a real performer.
How the rest of the music industry is responding
Spotify is not the only streaming company grappling with the growth of AI-generated music. Platforms are taking different approaches, ranging from disclosure and detection to outright restrictions.
The policies below reflect announcements available as of October 2026.
– Deezer: The streaming service reported on April 20, 2026, that it was receiving nearly 75,000 fully AI-generated tracks per day, accounting for approximately 44% of daily uploads. It said that 85% of streams on fully AI-generated tracks were detected as fraudulent and demonetised in 2025. Deezer tags detected AI music and excludes it from algorithmic recommendations and editorial playlists.
– Bandcamp: The platform announced in January 2026 that it would prohibit music and audio generated wholly or substantially by AI. It also prohibited the use of AI tools to impersonate other artists or styles in ways that violate its existing rules. This is a stricter approach than Spotify’s.
– Apple Music: The service has introduced AI transparency tags through its industry partners, with requirements for disclosure when AI is used to create a material portion of content. The aim is to give listeners more information about how recordings are made.
– YouTube Music: In October 2026, the platform began introducing labels for music made primarily using generative AI, relying on disclosures from partners alongside detection measures.
The differences matter. Some platforms want to identify AI-generated music and make it transparent to listeners, while others are restricting its distribution or removing it from recommendations. Spotify’s approach is to permit AI-assisted creativity while concentrating its enforcement on harmful or deceptive uses.
The AI debate: innovation versus protecting musicians
The music industry’s AI debate is no longer simply about whether a computer can produce a convincing song. It is increasingly about ownership, consent, income, authenticity and the future of creative work.
Supporters argue that AI tools can make music production more accessible. Independent artists who cannot afford large studios, session musicians or extensive production teams may be able to experiment with arrangements, improve recordings and develop ideas at lower cost. Producers can use the technology to explore new sounds, while established musicians can incorporate AI into their creative processes without necessarily surrendering authorship or control.
However, critics warn that AI-generated music can overwhelm streaming catalogues, imitate recognisable voices and compete for listeners’ attention without the same investment of time and resources associated with traditional music-making.
There are also questions about the music used to train generative AI systems. Artists and rights holders have raised concerns about whether their recordings, lyrics and distinctive vocal styles have been used with appropriate permission or compensation.
The financial stakes extend beyond individual streaming payouts. A study commissioned by the International Confederation of Societies of Authors and Composers (CISAC) and conducted with research consultancy PMP Strategy projected that approximately 24% of creators’ revenues could be at risk by 2028 as generative AI expands. The report estimated that this could amount to €4 billion annually by that year. These are projections, not confirmed losses already suffered by musicians.
There is also a transparency problem. In a survey commissioned by Deezer and conducted by Ipsos in November 2025, involving 9,000 respondents across eight countries, 97% of participants could not distinguish fully AI-generated music from human-made music in a blind test. Eighty per cent agreed that entirely AI-generated music should be clearly labelled, while 73% of music-streaming users said they wanted to know when a service was recommending fully AI-generated music.
The findings suggest that the debate is not necessarily about banning technology. It is also about ensuring that listeners can make informed choices and that artists are not misrepresented or deprived of control over their identities.
What record labels want going forward
Record labels, publishers, distributors and collecting societies are increasingly focused on creating rules that allow technological innovation without undermining existing rights.
One direction is greater transparency. Spotify’s support for the DDEX disclosure standard reflects an effort to establish a common way for labels and distributors to identify AI contributions in music credits. Such a system could help platforms, rights holders and listeners understand whether AI was used for vocals, lyrics, instrumentation or production.
Another priority is consent. AI-generated replicas of an artist’s voice should not be used to mislead listeners or commercially exploit an identifiable performer without permission. Labels and artists also need clarity about licensing arrangements when AI tools are used to create music based on existing recordings or recognisable performances.
A third concern is the protection of royalties. Labels want streaming revenue to reward legitimate listening rather than automated plays, duplicate uploads or schemes designed to exploit payment systems. Better detection and stronger distributor checks could help reduce the financial incentives for fraudulent activity.
The industry is also debating how AI-generated music should be credited, marketed and considered for playlists, awards and charts. There is no single universal policy across all platforms, and Spotify’s approach should not be mistaken for an industry-wide ban.
For musicians, the direction of travel is toward more accurate metadata, clearer disclosure, stronger identity protections and greater scrutiny of suspicious activity. For labels, the challenge is to protect their catalogues and artists while deciding how to work with AI companies and which uses of the technology they are prepared to license.
What this means for South African artists
South Africa’s music industry includes independent musicians, producers, songwriters, labels and distributors who rely on digital platforms to reach audiences locally and internationally. As AI tools become more accessible, artists may find them useful for production and experimentation, but they will also need to understand the rights and platform rules attached to their releases.
Before uploading a song, artists should establish who owns the composition and master recording, whether any samples or voice models have been cleared, how AI contributions should be declared and whether their distributor supports the relevant credits. They should also check that their artist profiles and release metadata are correct.
Musicians whose voices or identities are being impersonated should gather the relevant release details and report the issue through Spotify’s official reporting process. Artists can submit a claim through Spotify’s legal form under the Publicity / Likeness category when they believe their voice has been cloned without authorisation.
Labels and managers should likewise review their release procedures rather than assuming that a distributor will automatically detect every problem. AI detection tools have limitations, and Spotify has acknowledged that no single signal is sufficient to identify every suspicious track.
Ultimately, the new measures do not mean that artists must abandon AI. They mean that using the technology responsibly, respecting other artists’ rights and supplying accurate release information are becoming increasingly important.
As Spotify continues to refine its policies, the central question for the music industry remains whether it can encourage new forms of creativity without allowing mass-produced content, unauthorised imitation and streaming fraud to undermine the musicians whose work built the industry in the first place.
