Editor

Editor scores 8/100 on this site's structural exposure measure, which averages two independent task-exposure studies: OpenAI's "GPTs are GPTs" ratings and the AIOE index. A lower score means more exposed, and 8 is near the bottom of the range. The usage data points the same way. Microsoft's analysis of real Copilot conversations gives editors an AI applicability score of 0. 367, well above the 0. 159 average across the 785 occupations it covers.

The Anthropic Economic Index shows Claude use tied to editors' tasks, but at low volume, so we treat it as a directional signal only.

The shape of the work explains the score. Detecting errors in spelling, punctuation and syntax, rewriting copy to read more clearly, drafting headlines, checking text against a style guide and summarizing a long piece are all language tasks, which is where language models are strongest. The New York Times is a documented example of the direction of travel: leaked code showed an internal "OpenAI Styleguide" tool that checked headlines, bylines and copy against house style, plus a headline generator.

The Times said the project was a very early experiment that was not taken beyond testing and was not used by the newsroom, so it shows capability being tried, not editors replaced.

No legal or physical wall protects the core task. BLS lists no license or certification for editors, and the work is done at a desk. We found no statute that requires a human to edit a publication. What exists is contractual and narrow.

In a decision the PEN Guild announced on December 1, 2025, an arbitrator ruled that Politico violated its contract with the PEN Guild by launching two AI products, Live Summaries and the Capitol AI Report-Builder, without the required 60 days' notice and without human oversight. The arbitrator found no human editing of the summaries and Stylebook violations left uncorrected. That protects union members at one employer and does not bind the industry.

Policy is stronger than law here. The Associated Press's generative AI standards, published in August 2023, say AP staff do not use ChatGPT to create publishable content and that any output from a generative AI tool should be treated as unvetted source material. Poynter's account of the guidance adds AP's example that an AI-generated summary must be edited like any other content before publication. Read that two ways. It makes the human editor the accountable gate.

Our inference is that the job tilts toward checking machine output, and a smaller newsroom can staff that gate with fewer editors.

Direct evidence of lost editing work is thinner than the score suggests, and it should be stated plainly. A Washington University study of the Upwork freelance marketplace, covering writing and related fields including proofreading and copy editing, found that after ChatGPT's release monthly jobs fell 2% and monthly earnings fell 5. 2%. It is a short-window working paper.

A Globe and Mail report from November 2024 quotes a content agency telling one freelancer it had seen a decrease in writing and editing assignments. That is one anecdote. We searched for a named employer that replaced its editors with AI and found none.

Staff cuts to editors are real, but the sources we read do not attribute them to AI. The Texas Tribune laid off its entire copy desk in August 2023, with its CEO citing an unsteady economy, an evolving media industry and "the pressures of technology", and the Los Angeles Times cut 73 staff that June, about a third in news and copy editing.

In late August 2026 Lee Enterprises laid off editors and copy editors in Montana, Virginia and North Carolina. Poynter's report does not mention AI. A company spokesperson cited its cost structure, and the union at the Richmond Times-Dispatch said the company was prioritizing content producers over copy editors. AI may be part of the background, but the sources we read do not say so.

BLS projects employment of editors to fall 1. 1% from 2025 to 2035, from 105,000 to 103,800 jobs (the Occupational Outlook Handbook rounds this to 1%), against 3. 5% growth for all occupations. About 7,900 openings a year are expected, all from workers who transfer out or leave the labor force.

BLS's Job Outlook section for editors says that as traditional print newspapers and magazines lose ground to other media formats, employment of editors who work for those publications is projected to decline, while demand for reading for pleasure may support book editors.

Our reading is that pressure lands first on the proofreading and copy-level rungs, and on the entry roles from which many editors start. Executive and managing editors choose coverage, hire, set budgets and answer for what runs. Practitioners quoted in Publishers Weekly in late 2025 argue that developmental and stylistic editing resist automation. Those are opinions, and this site's checklist does not count "judgment" or "managing people" as protection without a legal or physical wall.

If copy-level work shrinks, fewer people may do the senior work, not more.

Two limits apply. Both structural inputs rate task content, not hiring, and a published score cannot show tools released after its data were collected. A single reader's job can also sit well above or below the occupation average, since a specialist editor of legal or scientific material is not doing the same work as a general news copy editor.