Trevor Paglen

Trevor Paglen is the geographer, artist, and author whose practice makes surveillance infrastructure, AI training data, and state power physically and publicly visible — interdisciplinary work across photography, sculpture, investigative journalism, and writing that ranges from long-lens photography of classified NSA installations to interactive AI installations at Fondazione Prada to books on machine vision for general readers (see Person entry). He is tracked here as a Voice because his sustained public-facing output — the surveillance-photography series that transformed the aesthetics of counter-surveillance disclosure; ImageNet Roulette (with Kate Crawford, 2019), the interactive installation that prompted ImageNet to remove over 600,000 images from its people categories; the Excavating AI essay (with Crawford, 2019) establishing dataset-politics as an AI-critique register; and How to See Like a Machine: Images After AI (Verso, 2026) — collectively engages non-AI publics with the argument that machine-vision systems are not neutral technical processes but power infrastructures, and that their operations must be made publicly visible before governance can begin.

Voice anchor

Paglen anchors the surveillance-and-AI-critique artist register — the corpus's voice slot for an individual whose practice engages general publics with AI critique through participatory installations, museum-scale exhibitions, and investigative disclosure rather than exclusively academic publication or organizational advocacy.

The existing corpus voices in the algorithmic-accountability and AI-critique register work primarily through academic publications and coalition organizing: Kate Crawford (Atlas of AI, the Knowing Machines Project, museum-scale art-research); Joy Buolamwini (Algorithmic Justice League, Gender Shades, facial-recognition bias research); Timnit Gebru (DAIR, Stochastic Parrots, LLM critique). None of these occupies the register Paglen anchors: the counter-surveillance investigative-artist making classified and proprietary systems physically tangible for museum-going and general-internet publics who encounter them not as policy abstractions but as visible, experiential artifacts.

Paglen's entry fills a specific gap: the corpus had no Voice carrying the participatory dataset-critique shape — an artwork that places the public audience inside the machine's judgment and produces a concrete documented outcome (dataset changes, civil-liberties recognition, movement uptake) — nor a voice anchoring the surveillance-disclosure photographer register where the physical infrastructure of state power is rendered as landscape art and then examined alongside the AI systems that surveil.

Core practice and signature works

The surveillance photography series — making the invisible visible. Paglen spent more than a decade photographing the physical infrastructure of US state surveillance: classified NSA bases in Maryland and the UK photographed using high-end optical systems, spy satellites tracked across night skies, undersea transoceanic cables where top-secret documents showed NSA tapping — work that required Paglen to learn scuba diving and underwater navigation — and classified spacecraft in Earth's orbit. The series reformulated the American landscape-photography tradition (Timothy O'Sullivan, Ansel Adams) as counter-surveillance disclosure: the same horizon, now revealing the surveillance state embedded in it. This body of work was recognized with the Electronic Frontier Foundation Pioneer Award in 2014 — awarded alongside former UN Special Rapporteur Frank La Rue and Rep. Zoe Lofgren — for helping "the world understand how technology and civil liberties are interwoven into our lives," and with the Deutsche Börse Photography Foundation Prize in 2016. The mid-career survey Trevor Paglen: Sites Unseen at the Smithsonian American Art Museum (June 2018–January 2019; Museum of Contemporary Art San Diego, February–June 2019) brought this work into full institutional framing.

Citizenfound (2014) — documentary reach. Paglen contributed research and cinematography to Citizenfound, Laura Poitras's Academy Award-winning documentary on Edward Snowden and the NSA surveillance programs — the highest-reach public artifact to which his surveillance practice directly contributed, extending his counter-surveillance disclosure work into mainstream cinema audiences far beyond the gallery circuit.

ImageNet Roulette (2019) — participatory dataset critique with documented outcomes. With Kate Crawford, Paglen created ImageNet Roulette, an interactive installation and web application where users upload photos and receive the AI labels drawn from the 2,833 subcategories of people in ImageNet. The results exposed racist, misogynistic, and degrading categorizations embedded in the dataset — labels including racial slurs, criminal designations, and identity categories rooted in nineteenth-century pseudoscience applied to contemporary photographs. When users posted labeled portraits online, the resulting public attention prompted ImageNet to remove over 600,000 images from its people categories — a concrete dataset-politics outcome and the exemplar case cited in the corpus's 2026-09-01 scope ruling on AI-critical artists. Paglen described the project's purpose as calling "attention to the real harms that machine learning systems can perpetuate".

Excavating AI (2019, with Kate Crawford) — the dataset-politics essay. The companion essay "Excavating AI: The Politics of Images in Machine Learning Training Sets" (published by the AI Now Institute, NYU, September 2019) conducted what the authors described as "an archeology of datasets" — examining ImageNet, JAFFE, UTKFace, and IBM's Diversity in Faces across taxonomies, individual categories, and labeled images. The essay demonstrated that image-classification in AI is "an inherently social and political project" embedding historical discrimination and pseudoscientific racial classification into systems shaping hiring, education, law enforcement, and healthcare. Published as companion to the Training Humans exhibition at Fondazione Prada (September 2019–February 2020) and From Apple to Anomaly at the Barbican Centre (London), the essay became a widely-cited anchor for the argument that training-data politics is a civil-rights and justice issue.

How to See Like a Machine: Images After AI (Verso, 2026) — the public-literacy book. Paglen's sixth book, published by Verso in May 2026, examines how computer vision and machine learning have restructured our relationship with images — "the vast majority of images are now made by machines, for other machines." Drawing on domains including psychological operations, UFO photography, stage magic, and PR strategy, the book teaches readers to ask not what images "say" but what they "do" and where they come from. Hal Foster describes it as "the toolkit we need" for understanding how computer vision and AI have restructured perception, labor, and reality itself. The Art Newspaper reviewed it as a "timely take on technology's hijacking of visual culture" in August 2026.

Public output and engagement

Paglen's public output runs through four overlapping channels, each reaching a structurally distinct audience layer.

• Museum and gallery circuit. Solo surveys at the Smithsonian American Art Museum (2018–19), Carnegie Museum of Art, Fondazione Prada, Barbican Centre, and Vienna Secession, alongside group exhibitions at the Metropolitan Museum of Art, SFMOMA, and Tate Modern. The Smithsonian survey was the first institutional framing of his counter-surveillance practice as a public-interest civic discourse alongside its art-world reception.
• Participatory digital installations. ImageNet Roulette functioned as both a gallery installation and a publicly accessible web application, reaching mass internet audiences who uploaded their own photos and shared AI-generated labels — a participatory-public-engagement shape absent from every other voice in the corpus.
• Investigative journalism and documentary. His contribution to Citizenfound (2014) and photographic documentation of NSA surveillance infrastructure operate in the investigative-journalism register; his EFF Pioneer Award situates him within the civil-liberties and digital-rights movement, not only the contemporary art world.
• Books and essays for general audiences. From experimental-geography studies (Blank Spots on the Map, 2009; I Could Tell You But Then You Would Have to Be Destroyed by Me, 2007) through the dataset-politics essay (Excavating AI, 2019) to the general-reader public-literacy book (How to See Like a Machine, 2026) — a sustained writing channel alongside the visual practice.

Why this is a Voice entry

A Voice entry is warranted — rather than additional structure on the Person entry — because Paglen's public-facing output is itself the load-bearing object the corpus needs to track: ImageNet Roulette and its documented outcome of 600,000 images removed from ImageNet; the Excavating AI essay establishing training-data politics as a civil-rights issue alongside the Training Humans exhibition reaching Fondazione Prada audiences; the surveillance-photography series recognized by the EFF as a contribution to civil-liberties understanding; Citizenfound's cinematic reach; and How to See Like a Machine as the current public-register output for the argument that machine-vision systems require public accountability. The corpus's AI-critical voices had covered the algorithmic-bias-and-harm register and the extractive-AI-critique register but had no voice carrying the counter-surveillance investigative-artist and participatory-dataset-critique register — the shape of engaging non-AI publics by placing them directly inside the machine's judgment and making the judgment's consequences tangible through documented outcomes. Paglen's entry closes that gap. Biographical detail and affiliations are recorded on the linked Person entry per the corpus's Person/Voice split.