Why Can’t This Just Work?: What Music Tech Needs to Move Forward

An AFEM Insight Report | Date: March 2026

Written by: Rufy Ghazi, AFEM Executive Board Member, Technology & Software

---

Executive Summary

Music technology companies have never been better placed to build the infrastructure the industry needs. The tools, the ideas, and the engineering capability to modernise it already exist. What the industry needs is a structural environment that enables this innovation: consistent metadata enforced at the point of creation, clear legal frameworks for AI, and rights infrastructure built for fairness and collaboration. AFEM surveyed 22 companies across the music technology ecosystem in early 2026 to understand, in specific terms, what that environment would look like.

The metadata problem is the most foundational. The opportunity lies in consistently applying the standards that already exist. On AI, legal frameworks are still taking shape across most major markets. The more immediate challenge is cultural: the reflex of treating all AI as a threat and making no distinction between GenAI tools and those built to identify, attribute, and promote human-made music. Clarity and nuance, in equal measure, could open significant ground.

Rights fragmentation, the absence of unified APIs, and the gap between licensing timelines and tech development cycles each add to the overhead. Progress is already underway on each of these fronts. ISNI adoption is gaining real traction, DDEX enforcement is generating the kind of behavioural change that years of encouragement couldn’t, and platform-level action on AI disclosure is gathering momentum. The coordination and collective will to use what is already there: that is what this report asks for.

The Question That Keeps Coming Back

“Why can’t this just work?” It surfaces constantly across the music technology ecosystem as a genuine expression of frustration from companies that have the engineering capability to build remarkable things. But these companies regularly find that the current structural conditions of the music ecosystem make innovation and related integrations difficult.

The Metadata Problem

A track’s journey from studio to streaming platform involves dozens of handoffs, and at each one, the information describing it can be left incomplete or duplicated. This is the metadata problem, and it lies beneath almost every other infrastructure challenge in our industry. Half of survey respondents, 11 out of 22 companies, identified conflicting metadata across databases as their single biggest structural challenge.

Chloe Dagenais, founder of Music Team, a platform combining catalogue management, distribution, and rights registration, puts it plainly: “To me, the conflicting metadata is a symptom. And the core issue is the fact that there’s no mandatory metadata at the point of creation.”

The industry has developed a habit of releasing music first and reconciling data later. Manual entry across systems multiplies errors at every stage. A track moves from producer to distributor to DSP to collection society, and each handoff is a potential point of failure. Artists and producers often discover the consequences only when royalties do not arrive. The Fair Play report on UK electronic music royalties, published in November 2025, put a number on the scale of that problem: only 36% of electronic music performances result in the correct creator receiving payment, with an estimated £5.7 million misallocated annually in nightclubs alone. (1)

41% of survey respondents also cited the absence of universal artist and song identifiers as a core problem. The International Standard Name Identifier (ISNI), which assigns a unique code to each contributor to creative works, has gained meaningful traction recently. Universal Music Group embedded ISNI into its A&R and supply chain systems in early 2025, assigning identifiers to over 100,000 contributors. Spotify, Amazon, Meta, and SiriusXM have since integrated ISNI into their processes, such as reconciling name variants. This is a signal that appetite for standardisation exists when there is a good solution, a clear pathway for its adoption and industry pressure.

The mid-year report on ISNI adoption published in August last year (2) is honest about the work that still needs to be done: identifiers need to be captured at the point of creation rather than retrofitted, and real pilots between labels, publishers, DSPs, and collection societies to be tested so the practicality can be tested end-to-end. 

A co-founder at a music tech company added an observation that highlights another factor driving the metadata problem:
“This complexity and this intransparency also benefits players in the music industry. We have to understand how we can incentivise those.”

To put it mildly, the argument is that the key players who control metadata lack motivation to share information in the absence of incentives. This is the structural tension underneath the technical problem.

The past shows why. The Global Repertoire Database (GRD) spent years and considerable investment attempting to solve exactly this problem before collapsing in 2014, when Performing Rights Organisations could not agree on governance and billing rules. The lesson was simple and clear: institutions protect what gives them leverage. Getting data right at the point a track is created, before it ever enters the system, is a more realistic place to start. With AI in the picture, the metadata problem is now being built into every AI tool that identifies, attributes, and recommends music. Hence, the cost and repercussions of delay in fixing this problem are rising rapidly.

The AI Framework Problem

As we know, the legal frameworks that govern how AI companies can use copyrighted music are either absent or still being written. For the companies working with AI, that gap has real operational consequences. In our survey, 27% cited unclear legal frameworks for AI and emerging technologies as a major barrier. Among companies actively working with AI, there are specific challenges: just over a third lack a reliable way to document and verify which copyrighted works were used to train models. The same proportion also reports the absence of a fixed copyright framework for AI-generated content. An additional 29% encounter reluctance from rights holders to engage before legal precedents are established, while navigating platform policies on AI content that keep shifting without warning.

One of the survey respondents shared a potential practical consequence:

“Slow responsiveness of rights holders for sync requests will drive potential users towards AI-generated music for their productions.”

When delays intended to protect rights make licensed music harder to access than synthetic options, the market finds a way around the problem. The lack of a framework is worsening the problem, all the while players like Suno continue to grow. The company, currently in active litigation with Universal Music Group and Sony Music Entertainment, announced $300 million in annual recurring revenue and over 100 million users (3).  The longer a workable framework takes to arrive, the less leverage the rights side will have when it does.

There is a more nuanced problem underneath the general legal uncertainty. Roman Gebhardt, Chief Artificial Intelligence Officer at Cyanite, a company using AI for music search, recommendation, and licensing enablement, described it from his experience:

“There is this general reservation towards AI. If that term comes up in the music field, everyone directly jumps onto the opt-out button.”

The difficulty is that “AI” covers a very wide range of applications. Generative tools that produce synthetic music from existing training data sit in a fundamentally different category from tools that analyse, identify, and attribute human-made music. As Roman observes, the word alone is enough to trigger a refusal, regardless of what the tool actually does.

AFEM’s own AI Principles set out what responsible practice should look like: explicit consent before training, transparent documentation of which works were used, fair remuneration for both training and generated outputs, and protection of moral rights. Survey respondents were asked which of these principles is hardest to put into practice. The most common answers were consent mechanisms and fair remuneration models. The challenge is operational: the infrastructure to execute them simply does not exist yet.

Some recent developments on the platform side are encouraging. Spotify introduced mandatory AI disclosure fields via DDEX in September 2025 (4), with non-compliance resulting in removal or reduced visibility. Deezer went further: it has tagged over 13 million AI-generated tracks, demonetised up to 85% of all streams on AI-generated music, and is now licensing its detection technology to others (5). Apple Music followed recently with mandatory Transparency Tags, requiring labels and distributors to disclose AI use across four categories: artwork, sound recordings, lyrics, and music videos(6). Even though positive developments, these distinct philosophies across platforms will create complexity for distributors, labels, and artists navigating platform rules. This yet again reinforces the case for industry-wide standards.

Three Further Pressures

Other major barriers include rights fragmentation, the lack of unified APIs, and a mismatch in speed between tech development and rights. Rights fragmentation shows how music ownership is organised: a single track can have different rights for the master recording, the underlying composition, and public performance. All of these rights are held by different stakeholders across more than 120 collection societies worldwide, with variations depending on the territory. The consequences? Music tech companies pulling back from geographic expansion, narrowing product scope, or walking away from markets that are complex and unviable.

Philipp Köhn, co-founder of twelve x twelve, a company providing financing to music rights holders, drew a comparison worth sitting with, highlighting that the music industry has built no comparable infrastructure for rights data.

“If we look at standards, financial industries have SWIFT, an international standard that everybody can use. When you look at real estate, land registries get financed because every time a transaction happens, a tiny amount of money gets paid.”

The absence of unified APIs adds another layer of complexity. Without open interfaces between rights databases, distribution platforms, and catalogue systems, companies build expensive custom integrations for every connection. The speed mismatch sits across all of the above. Tech companies develop on weekly or monthly sprint cycles. Licensing processes can take months. And as you would imagine, these gaps lead to resources being deployed to build workarounds.

The Cost of Complexity

When navigating the system becomes harder than building the product, it signals structural friction in the ecosystem. The overhead of navigating the ecosystem’s fragmented rights infrastructure consumes engineering capacity that would otherwise go into building: tracking down rights holders, reconciling conflicting metadata, managing every jurisdiction separately. Half of the companies that took the survey said that more than half of their total development resources go towards workarounds rather than core product innovation. The consequences are predictable: delayed launches, pivots away from original visions, and markets left on the table because the rights landscape made them legally unnavigable. When the cost of compliance consistently outweighs the cost of building, these outcomes follow.

It is worth pointing out that these costs do not sit only with the companies building the tools. Every tool that does not get built, every feature that gets descoped, every market that gets abandoned is a gap in what the industry could offer artists, rights holders, and audiences.

What It Would Take

What would help? Ask the companies building in this space, and the answers come back consistent: shared infrastructure, legal clarity on AI, incentives to drive standards adoption, and genuine interoperability across platforms. These are requests for the operational conditions that allow legitimate innovation to function, all the while respecting the system. As one survey respondent put it:

“This would remove friction, reduce legal uncertainty, and allow innovators to focus on building products instead of navigating fragmented systems.”

The webinar discussion revealed something worth noting. The panellists, all founders of music technology companies, agreed that the technical solutions to most of these asks already exist. What is harder is the coordination, and underneath coordination, the incentives to change. Chloe Dagenais identified the gap precisely:

“We’ve got data standards. But no one is enforcing those standards. It feels like somebody needs to put a bit more pressure on the system.”

DDEX, the Digital Data Exchange, has published data communication standards for the music industry for years. What has been missing is any mechanism to drive uptake of the existing standards. When Spotify attached real consequences to DDEX AI disclosure fields in September 2025, 15 major distributors and labels committed within weeks.

On the question of where to begin with infrastructure, Philipp Köhn offered a practical framing: rather than trying to retrofit standards onto decades of legacy catalogues, establish what good looks like for everything created from this point forward. He also pointed to AI’s positive role in that process.

“The task of cleaning up the past is absolutely overwhelming. But we could start to say, let’s come together and just start something new. All the intellectual, redundant, tiny lab work that needs to be done to clean this up is something perfectly suited for AI.”

It is clear that what is being asked for is already within reach, and progress on several fronts is already underway. Platform enforcement is creating the behavioural change, identifier adoption is gaining traction, and political attention on AI licensing is intensifying. The ask from the tech sector is for that progress to become deliberate rather than incidental.

The Way Forward

These barriers have been part of the conversation for years. The ask to rights holders, collection societies, and policymakers is specific: to understand the conditions that would allow innovation to happen within a fair, rights-respecting system, and to consider whether those conditions are worth creating. AFEM’s role is to bring the different parts of the ecosystem into the same conversation. The aim has been to give that dialogue more substance and less room for ambiguity about what would actually help. It’s evident that the infrastructure to build something better exists. Whether the ecosystem chooses to act on it is the only question that remains.

---

This report was produced by Rufy Ghazi in her capacity as AFEM Executive Board Member for Technology and Software, drawing on survey responses from 22 music technology companies, discussion from AFEM’s Q1 2026 webinar, and desk research from industry publications. Survey respondents and webinar participants are not individually attributed in line with the session’s safe-space agreement. Roundtable quotes are reproduced with participant consent.