When Idaho voters read an op‑ed signed by a state official, they expect a human voice. Recent testing by Idaho Education News, however, suggests that many of those pieces were at least partially generated by artificial‑intelligence tools such as ChatGPT, Claude and Gemini.
AI detectors spot AI‑written content in dozens of statements
Using two commercial detectors, Pangram and GPTZero, EdNews identified AI‑generated text in op‑eds and statements from several high‑ranking officials. Gov. Brad Little’s office produced three pieces flagged by the tools; Attorney General Raúl Labrador’s office also had three, while State Superintendent of Public Instruction Debbie Critchfield’s office had two.
None of the offices disclosed that they employed AI in the drafting process. The detectors classified the writings into three categories: entirely AI‑written, AI‑assisted, or human‑written. For example, Pangram determined that an op‑ed by the governor on nuclear energy was 100% AI‑generated, while another praising Idaho’s fiscal discipline was 88% AI‑written.
Officials acknowledge limited AI use
When approached for comment, the governor’s communications director, Emily Callihan, said, “AI did not write the three op‑eds you noted, but AI has replaced the use of other writing resources, such as an online thesaurus and editing tools.” Labrador’s communications director, Daniel Estes, admitted that he and director of external affairs Damon Sidur use AI to clarify complex legal language, but insisted that the final pieces are edited and substantially revised by humans.
Critchfield’s spokeswoman denied any AI involvement, while the superintendent’s communications director, Andrea Dearden, provided draft versions of the op‑eds to demonstrate human authorship. Nonetheless, Pangram’s analysis showed that a later draft of one piece was 74% AI‑written after significant edits.
Reliability of AI detectors questioned
Both Pangram and GPTZero acknowledge that AI detectors are not infallible. Pangram reports a false‑positive rate of 0.0041%, roughly one in 24,000 documents, and an independent University of Chicago study found “near‑zero” error rates for medium‑ to long‑length texts. Yet the tools can misclassify historic documents; earlier versions mistakenly flagged the U.S. Constitution as AI‑generated.
Critics of the detectors argue that the technology is still evolving and that reliance on such tools could lead to unwarranted accusations. Nonetheless, the Idaho Education News investigation highlights a growing need for transparency when public officials employ AI to shape public discourse.
What this means for Idaho voters
Idaho’s elected leaders routinely task staffers with drafting speeches, social‑media posts and op‑eds. The emergence of AI as a writing aid raises legitimate questions about authenticity and accountability. Voters deserve to know when a piece of communication is primarily the product of a machine versus a human mind.
As the state continues to explore the benefits of AI in government, officials may consider adopting clear disclosure policies. Such steps would reinforce trust, uphold the principle of transparent governance, and ensure that Idaho’s citizens can evaluate the ideas presented on their own merits.
Original reporting: Idaho Education News — read the source article.
