Amongst the key topics on the minds of music business leaders this year are AI and streaming fraud.
Sony Music Group Chairman Rob Stringer for example, during Sony Group Corporation’s annual presentation to investors on Wednesday May 24, said that, “Fraud on key DSPs is a problem that must be eliminated through aggressive enforcement by these DSPs and distributors, or by changing payment methods [i.e. royalty payout models] to better reduce the incentive for fraud”.
Stringer’s comments arrived in the same month that Spotify removed a substantial number of tracks – many created via AI music-making platform Boomy – from its service, citing “potential cases of stream manipulation”.
Via generative AI music apps, large volumes of audio content can be created by fraudsters and uploaded to DSPs with the aim of racking up huge numbers of plays of this content via bot-driven ‘streaming farms’.
Back in January, we reported on a recent French study showing that up to 3% of music streams on services like Spotify are known to be fraudulent.
As previously noted by MBW, this number only represents the ‘fake streams’ that services can actually detect; it doesn’t include the ones they don’t find.
Generative AI also presents challenges for rightsholders and DSPs today, from proliferating the vast daily supply of content to platforms, to AI models trained on copyrighted material and the legalities around the replication of superstar artists’ voices, as witnessed earlier this year with the ‘Fake Drake’ controversy.
On Tuesday (June 6), France-born music streaming service Deezer set out a strategy to address both the rise of AI music and fraudulent streaming activity on its platform.
Deezer says it is building a set of tools to detect AI-generated content on its platform, and aims to develop a system for tagging music that has been created by generative AI, starting with songs using synthetic voices of existing artists.
The tags will be used to keep artists, labels, and users informed about what’s “real” or AI–generated on the platform.
Deezer claims that its plan will reduce fraudulent activity, and that it also plans to “develop a remuneration model that distinguishes between different types of music creation”.
Deezer tells MBW that, in 2022, around 7% of streams on the platform were detected as fraudulent.
The company says that its new toolset leverages years of research on Audio AI and content identification. Dezer’s proprietary “Radar” technology serves as the foundation for its fraudulent content detection.
‘Radar’ can scan large catalogs of music, identifying any song even when the signal is distorted, tempo changed, etc.
“Our goal is to weed out illegal and fraudulent content, increase transparency, and develop a new remuneration system where professional artists are rewarded for creating valuable content.”
Jeronimo Folgueira, Deezer
“With over 100,000 new tracks uploaded per day to our platform, it’s becoming increasingly important to prioritize quality over quantity and defend real artists that create truly valuable content,” said Jeronimo Folgueira, CEO of Deezer, commenting on the company’s new AI strategy.
Folgueira added: “AI can be used to create new incredible content and I believe there are massive benefits of using generative AI, but we need to ensure it’s done in a responsible way.
“There’s an opportunity now to get things right from the start of the AI revolution, and not make the same mistakes as the social media giants did when fake news started to flood their platforms. We owe it to the artists and the fans.”
Deezer’s announcement this week follows remarks made about AI by Folgueira to analysts on the company’s Q1 earnings call in April 25, when he said that, “We want to give our customers a high-quality experience and relevant content, so obviously getting AI to flood our catalog is not something we’re super keen on, and we’re working on that.”
On that same call, however, Folgueira revealed that Deezer has itself used AI to generate content for its recently-launched wellbeing app, Zen by Deezer, which offers music and audio content to aid sleep, relaxation and meditation.
“We do see the benefits of AI in terms of generating some kinds of content, especially at a very low cost,” Folgueira told analysts on April 25. “We have actually used AI to generate content for Zen, our new wellbeing app.
“Some of the content has been generated by AI at a very, very low cost. It’s a model where we own the content and don’t have to pay [outside content creators].
He added: “So we have actually used the benefits of AI to create a super-profitable model when it comes to Zen.” He also noted however that Deezer has “not explored taking that [AI content] in any shape or form into our music core business yet and currently it’s not our intention”.
“There’s an opportunity now to get things right from the start of the AI revolution, and not make the same mistakes as the social media giants did when fake news started to flood their platforms. We owe it to the artists and the fans.”
Jeronimo Folgueira, Deezer
Deezer’s new strategy around AI and fake streams follows the company’s confirmation in March 15 that it is working with Universal Music Group to investigate new economic models for music streaming.
According to a press release at the time, through this collaboration, “UMG and Deezer aim to develop new methods that holistically reward recording artists and songwriters for the value they create and to reimagine and update the engagement model for Deezer’s users and the artists they love”.
(Deezer launched a campaign to publicly champion a user-centric payment system (UCPS) back in September 2019).
“As a leading streaming platform, Deezer has a responsibility to create a fair and transparent environment for music consumption,” added Jeronimo Folgueira in a statement issued on Tuesday (June 6).
“Our goal is to weed out illegal and fraudulent content, increase transparency, and develop a new remuneration system where professional artists are rewarded for creating valuable content.”
He added: “This is why we have embraced the discussion around a new artist-centric model, and we are now also developing tools to detect AI-generated content.”
MBW asked Deezer a few questions about the company’s processes around stream manipulation and the rise of generative AI, answered here by Deezer’s Ludovic Pouilly, SVP Institutional and Music Industry Relations:
We reported on an investigation from the Centre National de Musique (CNM), which concluded that at least 1%-3% of music streams in France are fraudulent. What are your thoughts about these findings?
We can confirm these figures since we participated in the study and have shared our data with the CNM. However, these figures were based on 2021 and since then we have greatly improved our detection algorithm. In 2022, around 7% of streams were detected as fraudulent on Deezer.
Could you speak to the economic impact from streaming fraud for DSPs, artists and labels?
Streaming fraud has a real potential to impact labels and the opportunities for genuine artists to make a living out of their art and we started working on fraud detection over 10 years ago to make sure that artists are fairly remunerated.
Streaming fraud doesn’t have an economic impact on Deezer per se, since the amount of money we redistribute to labels and distributors is the same with or without fraud. But we do spend time and resources on developing our fraud detection algorithm, with 15 people working with fraud detection as part of their day to day. But it does impact artists and labels as they record a loss of revenues.
Do you have stats you could share about the growth or decline of streaming fraud on Deezer specifically?
We are constantly improving our detection algorithms and although there seems to be an increase in the past year, it’s difficult to determine whether this is due to us getting better at exposing fraud or if there’s actually more fraudulent activity.
In What markets is streaming fraud most prevalent in on Deezer?
We see fraud across all markets.
What is Deezer doing, and investing in, in order to fight and prevent streaming fraud on its platform?
Over the past 10 years, we have developed multiple algorithms to detect different kinds of fraud, and we currently have a team of 15 people working with fraud detection as part of their job role. We are continuously developing our algorithms through cutting-edge machine learning and AI.
Does tackling streaming fraud require a cross-industry collaboration to combat it, or should it be a DSP’s responsibility to fight fraudsters on their own platforms?
We can’t fight fraud alone and we need to collaborate across the industry, including labels and artists, to achieve the best results. Transparency is key and we are sharing as much info as we possibly can with our network.
How have the methods of artificially inflating listening figures evolved over the past few years and what are the most prominent methods today?
Fraudsters are constantly trying to game the system so their methods have evolved at the same pace as our algorithms. We are seeing smarter ways of “listening” and fraudulent behavior which is closer to reality. In the beginning, it was mostly loops and skips after 30 seconds to optimize the number of streams.
What are your predictions for where the music business will be in terms of the prevalence of streaming fraud over the next 5 years?
Streaming revenues are increasing year by year, which will most likely lead to an increase in fraud.
Is it possible to completely stop fake streams on streaming platforms?
There’s always been fraud and there always will be. Where there is money, there is going to be fraud. However, we are constantly working to improve our detection capabilities and we support any initiative to lower the incentive to commit fraud and make sanctions more severe.
We will continue to work with all our stakeholders to make it harder for anyone who wants to make money from fake streams.
MBW reported in MAy that Spotify had REMOVED a number of tracks created via AI music-making platform Boomy from its service because it detected artificial streaming of these tracks. Has Deezer removed any music from Boomy? Or any other AI music apps?
We don’t remove any tracks unless they’re related to fraudulent activity or include illegal, hateful or discriminatory content. AI-generated songs are not removed by default. We haven’t actively removed any content coming from Boomy, but the company itself removed content from Deezer.
How big of a challenge has generative AI been for Deezer in terms of its use in streaming fraud?
It doesn’t really change our process. Our algorithms are mainly based on listening behavior and not the content itself, and AI-generated music is treated the same way as any other content when it comes to fraud detection.
How concerned are you about the use of generative AI for artificial streaming?
We have a sophisticated fraud detection system, which is constantly being improved to handle any current or upcoming threats, including an increase in AI-generated content.
We are currently working on a next-generation AI detection tool to make sure we continue to deliver the best possible user experience on Deezer and protect the interests of artists and labels.
Music Business Worldwide
