From Piracy to Prompting: The Long Road That Made Creative Theft Feel Okay

In earlier eras of copying, a human was present at every stage. Copying required intention, judgment, and meaning-making. Even mechanical reproduction depended on someone deciding what to copy, how, and why. Modern Generative AI breaks that chain. A user issues an abstract command and receives output assembled from undisclosed sources by a system that offers no lineage, authorship, or accountability.

This is not simply a new form of piracy. It is the attenuation-often the effective removal of human aesthetic agency from the act of creation. Copyright law was never designed to regulate works where no accountable creative subject exists.

When a human decides to create art, they choose tools in service of an intention. Those tools are shaped by access, experience, economics, and inheritance. Some artists buy instruments; others inherit them; others improvise with what is available. Before pencil meets paper, hundreds of decisions have already occurred. Medium, limitation, taste, and history shape the work. These choices are not incidental, they are authorship in its most basic form.

Generative AI introduces a different structure. A prompt like “make a pop song with jazz influence” or “generate an alternative version of ‘It’s Oh So Quiet’ by Björk” does involve human intention, but only in a minimal sense: the selection of a target aesthetic. What the human does not determine are the musical decisions that traditionally constitute authorship: tonality, harmony, rhythm, structure, arrangement, phrasing. Those decisions are not interpreted by a performer but inferred statistically by a system trained on vast quantities of prior human labor.

This distinction matters. A score constrains interpretation; a prompt under-specifies outcome. A performer interprets within authorship; a generative model synthesizes without one. Delegation is not authorship. Selection is not composition.

There is no true historical analogue for this. Human collaboration is consensual, negotiated, and accountable. Even in collective art, credit and responsibility-however imperfect-remain possible. Generative AI collapses that structure. Even when training data is licensed or legally defensible, the artists whose work formed the substrate retain no control over how their labor is recombined, represented, or aesthetically redirected.

This is not merely a legal issue. It is a moral one.

Artists create to work through joy, grief, trauma, obsession, and curiosity. Releasing a work into the world is not consent for its emotional content to be extracted, abstracted, and redeployed indefinitely. Exposure is not forfeiture. Publication is not moral surrender. Generative AI treats emotional labor as inert material-something to be mined, normalized, and redeployed without regard for intention or consequence.

At this point, quality becomes irrelevant. A generated song may be compelling or forgettable; that is beside the point. Artistic value has never been determined solely by outcome. It emerges from the transfer of human will and effort into a medium, followed by recognition from the world. That recognition presumes someone to recognize.

Now consider a celebrated AI-generated song composed from statistical fragments of hundreds of human works. Who receives credit? What happens to the artists whose labor made that output possible? The listener encounters a finished object with no visible lineage-no names, no debts, no echoes to trace.

When humans make music, we hear history. We hear influence and inheritance. Generative AI collapses that chain. What it produces often resembles collage rather than composition-not because collage is illegitimate, but because collage traditionally names its sources. Here, the sources are obscured by design.

Genre itself begins to fracture. Genre is not just surface pattern; it is historical practice. Bo Diddley created the Bo Diddley beat, and it is attributed to him. If a model generates a variation in response to a prompt, who is the originator? If lineage cannot be traced, legacy becomes noise. Innovation becomes unmoored from its creators.

This system exists within a material reality that cannot be ignored. Generative AI is designed to reduce cost, eliminate risk, and displace labor. It centralizes value while diffusing authorship. A small number of institutions extract from a vast, uncredited creative commons and return polished artifacts stripped of history. This is not a neutral technological evolution; it is an economic reconfiguration with aesthetic consequences.

Many defenses of generative AI point to prompting as intention, tools as precedent, collaboration as analogy, influence as justification, or future attribution as a solution. Taken individually, these claims are not unserious. Taken together, they miss the point.

Tools extend judgment; they do not replace it. Collaboration presumes consent. Influence is interpretive; training is extractive. Lineage may be imperfect, but it is not optional. Ethics deferred are ethics denied. Responsibility dispersed until no one bears it is responsibility dissolved.

The moral failure of generative AI is not that it produces bad art. There’s no doubt it’s created works that have entertained the masses. I myself have been fooled by songs generated on Suno. I think the failure here is much greater. That failure is that it produces unaccountable art: output without agency, expression without responsibility, creation without lineage.

The question, then, is not whether generative AI is impressive, legal, or inevitable. The question is whether a culture willing to abandon authorship, lineage, and agency in exchange for frictionless output understands what it is relinquishing - and whether it is prepared to live with the consequences.

⸻

An invitation to you, the reader.

I’m interested in thoughtful disagreement as much as agreement.

If you work with generative tools, resist them, or feel conflicted somewhere in between, I’d genuinely like to hear how you’re thinking about authorship, responsibility, and lineage right now. Where do you see this argument holding and where do you think it breaks?