In March 2018, a Chelsea gallery sent out a press release for a mid-career painter’s solo show. Two consecutive paragraphs in that release had appeared, with minor word substitutions, in a different gallery’s release for a different artist on a different continent fourteen months earlier. The overlapping phrases were structural, not thematic — a construction comparing painterly gesture to “the residue of decision-making” and a clause about “the surface as both membrane and document.” Neither gallery acknowledged the overlap. Neither artist complained. No critic flagged it. The matter circulated among a handful of writers and editors for about a week, then dropped.
The disappearance tells you more than the plagiarism does. The text was not written by the named critic in either case. It started as talking points handed by one gallery director to a freelance writer, who shaped them into prose for roughly $400. The second gallery got a different press release from a different freelancer who had, at some point, absorbed the phrasing into a personal template. Nobody documented the chain of custody for the language because no art-world institution documents it. Catalogue essays, artist statements, wall texts, press releases, curatorial rationales — all circulate as though they emerge fully formed from the creative environment they describe. In practice they pass through a parallel authorship economy that has operated without acknowledgment for at least five decades.
The Siegelaub-Projansky Silence
The 1971 Siegelaub-Projansky agreement, formally titled “The Artist’s Reserved Rights Transfer and Sale Agreement,” remains the most consequential attempt to formalize the economic relationship between artists and the galleries that sell their work. Seth Siegelaub, a dealer and publisher associated with Conceptual art, and the lawyer Robert Projansky drafted a contract granting artists resale royalties, reproduction rights, and ongoing control over how sold works were exhibited and documented. The agreement circulated widely among artists and was adopted in fragmentary form by galleries in New York and Europe. Its influence on subsequent artist-dealer contract law — particularly in jurisdictions that enacted resale royalty statutes — is documented and traceable.
What the agreement never addressed was textual authorship. The contract specified who owned the physical object, who profited from its resale, and who controlled its reproduction. It said nothing about who wrote the text that accompanied the object into the world — the catalogue essay, the statement, the press release. That silence was not specific to Siegelaub and Projansky. It reflected a structural assumption shared across the dealer-artist-critic triangle of the period: critical writing was promotional labor, exchanged informally for access, advancement, or the vague promise of future assignments. Critics wrote essays for gallery catalogues without payment, without a contract, and without any claim to the text beyond a byline that could be declined at the gallery’s discretion. The gallery supplied the artist’s biography, the statement, the materials list, and the interpretive framing. The critic’s job was to elaborate, not to originate.
This arrangement produced a body of writing still cited as critical literature while having been generated under conditions closer to work-for-hire than independent assessment. The catalogues produced by Leo Castelli Gallery, Sonnabend Gallery, and Konrad Fischer Galgere in the late 1960s and early 1970s contain essays by critics whose names carry authority — Lucy Lippard, Germano Celant, Benjamin Buchloh — but whose drafts were subject to editorial control by the gallery and, in some cases, by the artist. The published text rarely reflects the full range of what the critic originally wrote. The cuts, additions, and substitutions made at the gallery’s request are not documented anywhere. The archive of correspondence between critics and galleries from this period, where it survives, reveals a consistent pattern: the gallery sends a packet of materials including a “suggested approach” or “key points to cover,” the critic drafts an essay, the gallery returns it with revisions, and the published version appears under the critic’s name.
The Access Economy
The mechanism that sustained this system was access. A critic who wrote unpaid catalogue essays for a gallery received first view of new work, introductions to artists, and the standing that came with publication in a gallery context. For critics building careers in the 1970s and 1980s, this exchange was not exploitative in any way they would have named. It was how the profession worked. The gallery needed text. The critic needed access. The artist needed interpretation. Everyone understood the transaction. No one recorded its terms.
That informal economy never dissolved. It professionalized into a tiered system that operates today with greater volume and less transparency. At the top tier, established critics are paid by major museums and large galleries to write catalogue essays, with fees ranging from $1,500 to $15,000 depending on institutional budget and the critic’s market value. These contracts typically include a kill fee, a deadline, and a word count, but they rarely include provisions for editorial transparency — no requirement that the gallery or museum disclose what changes were made to the critic’s draft, by whom, and on what authority. At the middle tier, freelance writers produce press releases, wall texts, and extended captions for fees between $200 and $800 per text, often working from talking points supplied by the gallery. At the bottom tier, interns and studio assistants draft artist statements that the artist approves without substantial revision, and that galleries then distribute as the artist’s own voice.
I have been on the receiving end of this system for two decades. Galleries have sent me pre-written artist statements with the request that I “adapt” them for publication. The statements arrive as Word documents on gallery letterhead, formatted as finished text. The implicit instruction is to paraphrase rather than originate — to produce a version that sounds like criticism while staying within the interpretive frame the gallery has established. When I have declined and written my own framing, the response from gallery directors has ranged from polite disappointment to the withdrawal of future access. The penalty for originating is exclusion. The reward for adapting is continued proximity to the work.
The 2018 Recycling Incident and What It Revealed
The 2018 incident of the recycled press-release language was not, in itself, significant. Two galleries used similar phrases to describe different artists. A few editors noticed the overlap and then forgot about it. What the incident revealed was the absence of any institutional mechanism for tracking the provenance of art-world language. The first gallery could not determine whether its talking points had been leaked, shared, or independently generated because it had no record of who wrote them. The freelancer who produced the first press release had moved on to other assignments. The gallery director who supplied the talking points could not remember whether they were original or borrowed from a previous project. The second gallery, confronted with the similarity, denied awareness of the first release and attributed the overlap to “convergent critical vocabulary.”
This is the structural condition the art world has created: a body of interpretive text with no documented chain of authorship, no editorial trail, and no institutional accountability for who wrote what, under what terms, and with what degree of autonomy. The same galleries that maintain meticulous condition reports for every painting in their inventory — documenting craquelure, retouching, and varnish thickness with photographic precision — cannot produce a single document identifying who wrote the press release for their last exhibition, what they were paid, and whether the text was edited after submission.
The AI Anxiety and the Misplaced Panic
The current anxiety about AI-generated art writing treats the arrival of machine-generated text as a crisis of authenticity in a field that has never been authentic about its text production. The Authors Guild, in its published AI Best Practices for Authors, identifies a genuine structural concern: that generative AI trained on unlicensed copyrighted work threatens to make human-authored writing “a rare luxury good representing only a minority of views.” The Guild’s advocacy for transparency, compensation, and disclosure in AI-assisted writing is a necessary intervention in publishing and literature. But its framework does not extend to the art world’s catalogue-essay economy, where the conditions the Guild warns about — uncredited labor, invisible editorial intervention, the displacement of original voices by institutional templates — have been standard practice for decades.
The panic over AI-generated press releases, artist statements, and catalogue essays assumes there was a prior state of transparent human authorship to be corrupted. There was not. The named critic on a catalogue essay may have written 30 percent of the published text. The artist statement on a gallery website may have been drafted by an intern and approved with a single read-through. The wall text in a museum may reflect the curator’s original language, or it may be a composite of edits by the communications department, the marketing team, and the artist’s dealer. AI does not introduce opacity into this system. The system was already opaque.
What AI introduces is scale and speed. A freelance writer who once produced four press releases per week can now produce twelve. A gallery that once commissioned one catalogue essay per show can commission three — one for the print catalogue, one for the website, one for the VIP preview mailer — at the same budget. The volume of interpretive text surrounding each exhibition increases, but the structural relationship between text, authorship, and accountability remains unchanged. For galleries and artists now exploring tools like AI novel writing software that supports structured editorial drafting, the appeal is not the replacement of human judgment but the industrialization of a workflow that was already semi-automated through templates, talking points, and adaptation requests.
The Parallel to Fabrication, Conservation, and Installation Labor
The art world has developed, over the past forty years, an increasingly precise vocabulary for crediting non-authorial labor. Major exhibitions now credit fabricators by name in wall text and catalogue. Conservation departments are acknowledged in loan agreements and acquisition records. Installation crews are listed in exhibition credits, and the preparators who handle specific works are documented in condition reports and installation logs. The 2014 exhibition “Zero: Countdown to Tomorrow, 1950s–60s” at the Guggenheim credited twelve fabricators and three conservators in its catalogue. The Whitney Museum’s 2017 retrospective of Hélio Oiticica named the conservation team responsible for reconstructing degraded Parangolés. These are not gestures of generosity. They are structural acknowledgments that the physical existence of art depends on labor that is not the artist’s own.
Textual labor deserves the same acknowledgment. The person who writes the catalogue essay, the press release, the wall text, and the artist statement is performing labor that determines how the work is understood, contextualized, and valued. The absence of their name from the credit system is not an oversight. It is a structural choice that preserves the fiction that interpretation emerges spontaneously from the work itself, rather than from a specific person sitting at a desk, receiving a packet of materials, and producing text under economic and editorial constraints they do not control.
The Broader Labor Pattern
The ghostwriting economy in the art world mirrors broader patterns of uncredited creative labor that economic data infrastructure tracks but cultural institutions do not. The Federal Reserve Bank of St. Louis maintains FRED Economic Data, a public database of labor-market time series including freelance employment, contingent work arrangements, and creative-sector compensation patterns. The data shows consistent growth in gig-economy and contract-labor categories across the period when the art world’s freelance writing economy expanded. The art world’s refusal to document its textual labor supply is not unique — it reflects a wider cultural refusal to treat interpretive and promotional writing as labor worth tracking. But the art world’s case is distinctive because the institutions involved are the same ones that maintain rigorous standards for documenting every other form of labor associated with art production.
A Structural Proposal
The solution is not disclosure of AI use. Disclosure of AI use in a system that never disclosed human ghostwriting is a fig leaf. The solution is documentation of all textual labor, regardless of origin. Every catalogue essay, press release, wall text, and artist statement should carry a credit line naming the writer, stating whether the text was edited after submission, and identifying the party responsible for the edits. This is not a radical proposal. It is the application of existing standards — the standards museums and galleries already apply to fabricators, conservators, and preparators — to a category of labor they have historically chosen to leave invisible.
The objection will come from galleries and museums that rely on the fiction of spontaneous interpretation to maintain control over how their exhibitions are framed. If the wall text is credited to a specific writer who was paid a specific fee and whose draft was edited by the communications department, the text loses its authority as institutional voice. It becomes what it has always been: a document produced under specific economic and editorial conditions, by a person with a specific relationship to the institution, subject to revision by parties whose interests are not identical to the artist’s or the public’s. That transparency is threatening because it exposes the machinery the art world depends on to present its products as autonomous cultural achievements rather than as the output of a coordinated production system.
The secondary objection will come from critics who fear that credit lines will reveal the extent to which their published work has been edited, shaped, or in some cases substantially rewritten by the institutions that commissioned it. This fear is legitimate. Transparency about editorial intervention will damage the reputations of critics whose bylines appear on text they did not substantially control. But that damage is the point. The current system protects critics by obscuring the degree to which their work is collaborative, and it protects institutions by obscuring the degree to which their “voice” is externally produced. Both forms of opacity serve the same function: they prevent the public from understanding how art-world language is actually made.
What the System Makes Possible
The catalogue essay industrial complex, as it currently operates, makes possible a specific kind of exhibition: one in which the interpretive framework appears to emerge from the work rather than from the institutional apparatus surrounding it. This appearance is not incidental to the art world’s functioning. It is foundational. The market value of contemporary art depends, in significant part, on the perception that the work generates its own critical context — that the artist is not only a maker of objects but a producer of meaning that critics merely observe and describe. If the public understood that the critical framing was typically supplied by the gallery, shaped by the marketing department, and written by a freelancer paid $400, the authority of that framing would collapse, and with it a portion of the work’s market value.
This is why the system persists. Not because anyone designed it, and not because anyone defends it explicitly, but because its opacity serves the economic interests of every party with the power to change it. The gallery gets interpretive control without paying for authorial independence. The critic gets a byline and access without disclosing the terms under which they wrote. The artist gets framing that flatters the work without acknowledging that the framing was manufactured. The museum gets wall text that sounds authoritative without accounting for the editorial process that produced it. AI does not threaten this arrangement. It makes it faster. The threat — and it is a real one — is that documentation of textual labor would make the arrangement visible, and visibility would make it accountable. That is the reform the art world needs, and it has nothing to do with whether the next press release is drafted by a human or a machine.