Current Affairs, the New Orleans-based nonprofit magazine, has adopted what it calls one of the strictest anti-AI policies in publishing as part of its new union contract. No AI-generated text may be published. No AI-generated art may be used as illustration. The policy is contractual, not aspirational — embedded in labor negotiations, not a press release. The move comes as mainstream publishing accelerates in the opposite direction. Business Insider's CEO has declared the publication "all in on AI." Sports Illustrated was caught publishing AI-written pieces under fabricated bylines. The Wall Street Journal defended running an AI-generated op-ed from hedge fund billionaire Stan Druckenmiller after it was exposed, on the grounds that it reflected his "genuine opinion." Guest op-eds in major newspapers have been found partly or wholly AI-written. The industry's gravitational pull is toward cheaper content at scale, quality and attribution be damned. The economic logic driving this shift is brutal and straightforward. Editor Nathan J. Robinson lays it out: if 100 AI posts cost near-zero and take four minutes versus paying a writer $5,000 for 20 posts, the spreadsheet math wins even if you lose 60 percent of your readership. For-profit publications operating under shareholder obligations have structural incentives to automate. Futurism reports more than 200,000 jobs in creative industries lost over four years, with game studios now running full AI art workflows. Pay rates and opportunities for artists are in freefall. Current Affairs occupies a genuinely unusual position in this landscape. As a unionized 501(c)(3), it has no shareholders demanding margin expansion and no profit motive to replace workers with machines. The magazine launched as a print publication in 2016 — against the advice of Vanity Fair's editor-in-chief — when every other outlet was pivoting to video. It has survived ten years and 62 print issues while better-funded ventures like the Messenger burned through $50 million and collapsed in a year. The policy does not ban all AI tool use outright. Writers may use AI-assisted research tools; artists may use tools for texture or color adjustments. The line is drawn at output, not workflow: the final product seen by readers must be human-generated. Robinson acknowledges this distinction while framing the core issue as whether technology serves workers or replaces them, citing Garrison Lovely's book "Obsolete: The AI Industry's Trillion Dollar Race to Replace Us" for the argument that AI developers are measuring success precisely by their ability to automate humans out of existence. The structural question is whether this model can sustain itself as competitors flood the zone with near-zero-cost content. Current Affairs is asking readers to vote with their wallets — subscribe, donate, pay for the handmade thing. The John Henry metaphor Robinson invokes is telling: inspirational, but the original John Henry dies at the end of the story. The magazine's survival depends on whether enough readers value provenance and craft to sustain a labor-intensive operation against an industry racing toward automated scale. What makes this interesting from a systems perspective is not the moral argument — most people agree slop is bad — but the institutional structure. A nonprofit with a union contract baking anti-AI provisions into labor law creates a different kind of binding commitment than a corporate policy memo that evaporates the moment a new CEO arrives. Whether that commitment is durable enough to matter at industry scale is the open question.
