Did Haley Stevens Use AI to Write Her Blog Post About Her Controversial Endorsement?

Last evening, Michigan Congresswoman Haley Stevens unveiled a blog post on X-formerly-Twitter under the title "The Decisions We Make." In the almost 900-word piece, Stevens detailed her controversial endorsement of Republican Mike Rogers for Michigan's vacant Senate seat, reversing her earlier backing of the Democratic contender, Abdul El-Sayed.

The race, one of the most contentious midterms in U.S. politics, could play a crucial role in tonight's Senate power struggle. The decision to withdraw support for El-Sayed has drawn harsh criticism from other Democrats, accusing Stevens of betrayal, while several Republicans, including President Donald Trump, have commended her for crossing party lines.

Stevens attempted to clarify her stance on CNN, but the interview reportedly backfired, with one commentator branding it "one of the most bizarre and incoherent political interviews I can recall." However, the blog post's impact may be even more significant. An AI detection tool, Pangram, analyzed the text and concluded that it was likely 90 percent generated by artificial intelligence.

The post demonstrates several telltale signs of AI-generated content, including a recurring "X dressed up as Y" phrasing and a reliance on the "rule of three," a common tactic in AI writing. The article also accuses Stevens' fiancé and others of influencing her decision, labeling such accusations as "sexism dressed up as political analysis."

Attempts to contact Stevens for comment have not yet yielded a response. The use of AI by politicians has come under increasing scrutiny as the technology becomes more prevalent in legislative circles. Earlier this year, The Washington Post reported that a significant portion of legislation is now being drafted using AI, often with little oversight.

Instances of AI-generated text appearing in official websites and speeches across the U.S. and abroad have also surfaced.