AI Is Everywhere in Galleries, a New Report Finds — But Few Have Rules for Using It
Artificial intelligence has slipped into the daily rhythm of commercial galleries with remarkable speed, and with surprisingly little oversight. A new “AI in Galleries” report from the art industry network First Thursday finds that 84 percent of galleries surveyed are already using AI tools as part of routine operations. Yet only 8 percent say they have a formal policy that sets boundaries for how those tools should be deployed.
The findings draw on interviews with 103 gallery professionals worldwide, including owners, directors, and staff, working primarily across the UK, Europe, and the US. Taken together, the responses describe an art trade adopting new technology quietly and pragmatically, while lagging on governance, strategy, and basic questions of accountability.
One of the report’s clearest signals is that AI use is often driven by employees rather than leadership. In most cases, staff are bringing tools into the workplace informally: 82 percent of respondents said they access AI through personal accounts, while just 18 percent reported being provided with dedicated workplace accounts.
That imbalance has immediate implications for data protection in a sector where confidentiality is a form of currency. Without clear internal guidelines, the report warns, staff may be entering sensitive information into platforms that galleries cannot supervise or control. The material at stake could include collector names, sales histories, and pricing discussions — details that can shape relationships and negotiations as much as they document them.
Despite the risks, AI is already being used across a broad range of tasks, with writing and editing at the center. Seventy-eight percent of respondents said they use AI to draft texts such as press releases or emails. More than half reported using it for translation, while others cited social media content, administrative work, market research, and researching collectors.
“The art market has mostly been silent on the use of A.I. within galleries, but it’s clear from our research that a silent transformation is underway,” Callum Halle-Thompson, First Thursday’s founder, said in a statement. He characterized the current moment as widespread experimentation rather than a coordinated digital shift, noting that fewer than one in 10 respondents reported having policies in place.
The report links this adoption to mounting operational pressure. Nearly four in five respondents said running a gallery has become more complicated over the past two years, pointing to rising administrative demands and the escalating costs of participating in art fairs. In that environment, AI is often less a visionary initiative than a time-saving patch — a way to offload repetitive tasks in teams that are already stretched.
Notably, fear of job loss does not appear to be the dominant mood. Only about 29 percent of respondents identified automation as a risk to employment. A more pervasive concern is quality control: around two-thirds said they worry about errors or low-grade content produced by AI, which can create new labor in the form of checking, revising, and verifying.
Even so, sentiment toward the technology skews optimistic. Sixty-two percent of respondents said AI adoption in the art market is inevitable, and more than half indicated they would invest in AI if the benefits were clear.
For now, many gallery professionals are relying on general-purpose platforms rather than tools designed specifically for the art trade. Respondents most frequently cited systems such as ChatGPT, Claude, and Gemini — a sign that the market for art-specific AI products has not yet displaced the convenience and familiarity of mainstream services.
The report’s underlying tension is straightforward: galleries are already using AI to write, translate, research, and streamline operations, but the infrastructure of responsibility has not kept pace. As AI becomes embedded in the mechanics of selling and communicating about art, the next competitive advantage may be less about adopting the tools than about setting standards for how — and on whose terms — they are used.
