A platform banning of AI and creative access

Ableism, AI, and the Politics of Access in Music Platforms

Recent moves by platforms such as Bandcamp to restrict or ban “AI-generated music” risk producing unintended but significant ableist exclusions. What is often framed as a defence of artistic authenticity or resistance to industrial automation collapses a wide spectrum of practices into a single undifferentiated category: “AI music.”

This flattening is not neutral. It obscures the difference between large-scale automated content production and the use of artificial intelligence as an assistive technology in creative work. For many disabled musicians, AI is not a shortcut or a threat to artistic integrity, but a necessary condition of access.

Artificial intelligence tools are increasingly used to reduce or bypass physical and neurodiverse or cognitive-related barriers in music production. They enable artists with limited motor control, speech, or traditional instrument access to compose and perform music through alternative interfaces such as text input, gesture tracking, or eye-movement systems. In these contexts, AI does not replace artistic agency – it enables it.

A blanket rejection of AI-generated work therefore risks reinforcing a narrow and implicitly able-bodied conception of creativity: one that assumes music must be produced through specific bodily capacities, workflows, and interfaces.

What appears as a critique of technology can easily become a gatekeeping mechanism that excludes forms of expression that depend on accessibility infrastructures.

The issue is not whether AI belongs in music. The issue is how platforms define legitimacy, and whether they are capable of distinguishing between industrial substitution and accessibility-driven artistic practice.

Without this distinction, “anti-AI” policies risk reproducing ableist exclusions under the language of ethical resistance. What is at stake is not technological purity, but participation. The question is not how to remove AI from creative spaces, but how to ensure that tools of access are not reclassified as forms of illegitimate authorship.

—

Alternative platforms and inclusive approaches

Beyond Bandcamp, there are already models that point toward more inclusive ways of handling AI-assisted music without resorting to blanket exclusion.

Platforms such as the Free Music Archive (FMA) may offer a more differentiated model. Instead of enforcing blanket bans on AI-generated content, FMA allows for structured attribution and licensing frameworks that can potentially accommodate AI-assisted works when properly declared.

Rather than treating “AI” as a categorical exclusion, such an approach makes it possible to distinguish between different production contexts, including assistive, accessibility-driven, and fully automated forms of generation. This shifts the focus away from prohibition toward transparency, accountability, and contextual tagging.

Within such a framework, AI use does not automatically determine legitimacy. Instead, it can be marked, described, and situated within clear authorship and rights structures—allowing for both ethical clarity and accessibility without exclusion.

Other platforms and infrastructures also suggest possible directions:

• SoundCloud allows more granular self-description and tagging, which can be used to indicate AI assistance without automatic exclusion.
• BandLab integrates digital production tools directly into its ecosystem, making it more open to hybrid human–machine workflows, including accessibility-driven creation.
• Internet Archive / Community Audio provides a non-commercial hosting structure where metadata and contextual framing can include experimental, assistive, or non-traditional production methods without strict stylistic gatekeeping.

More broadly, inclusive alternatives do not depend on a single platform, but on design principles:

• Mandatory but neutral disclosure systems (e.g. “AI-assisted / AI-generated / assistive AI used”) instead of bans
• Context-based labeling rather than categorical prohibition
• Accessibility-aware authorship models, where assistive technologies are explicitly recognized as part of creative agency
• Separation between industrial automation spam and assistive creative use, rather than treating them as identical
• Metadata-rich publishing infrastructures, where tools, interfaces, and constraints can be described without affecting eligibility

In this sense, inclusion is not about “accepting AI” in abstract terms, but about building infrastructures that can distinguish between extraction, automation, and accessibility.

Without such distinctions, platform policies risk reproducing a narrow idea of authorship that implicitly privileges non-disabled, tool-independent forms of production — and excludes precisely those creative practices where AI functions as an enabling condition of participation rather than a replacement of artistic agency.

—

An adjacent discussion

What matters in making music?

What counts as “music” is often less about the tools used and more about how decisions, listening, and shaping happen over time. If we take that seriously, then the question of AI in music cannot be reduced to whether something is “AI-generated” vs. “human-made.” It becomes a question of process, intentionality, and transformation.

A lot of music production is already deeply mediated: sampling, autotune, DAWs, algorithmic mastering, recommendation-driven aesthetics, and genre templates.

In that sense too, a 100% “pure” ideal of creation has never really existed as anything without a point of origin and continuity. By now the line between composition, editing, and computation is already blurred.

Given this perspective, AI can be understood as part of a continuum of computational and collaborative tools, rather than a rupture.

The key question is not whether AI is involved, but how it is involved:

• Is it replacing all decision-making in a fully automated pipeline?
• Or is it part of a dialogic process of selection, transformation, and reworking?

This is where the idea of deep collage becomes interesting. Collage as a method – cutting, layering, recombining, distorting, recontextualizing – has long been a legitimate artistic practice in music.

If AI is used to extend or intensify this logic, it can function as a tool for new forms of composition rather than as a substitute for it.

At the same time, listening practices matter too. Not everything produced with AI is slop simply because it involves AI. To be honest, much of today’s music is generic, repetitive, or aesthetically shallow – with or without the use of robotic intelligence involved.

The critical task is not automatic acceptance, but actual listening and judgment of results, rather than treating the production method as the primary criterion.

This also raises a deeper question: where do we draw the line between “artificial” and “non-artificial” in music at all? Rhythm quantization, synthesis, digital editing, and algorithmic structuring are already forms of artificial mediation. Computing is not an external intrusion into music – it is already one of its infrastructures. And some genres consciously base of these audible developments.

So the real issue is not artificiality versus authenticity, but which forms of mediation are acknowledged as creative, and which are disqualified in advance.

If we shift the focus this way, AI stops being a categorical threat or exception, and becomes part of a broader question of musical practice:

how sound is shaped through layered systems of human and robotic agency – and how we decide what counts as meaningful transformation within that process.

—