AI detectors like GPTZero fuel false cheating accusations

AI detectors—tools like GPTZero, Turnitin, and Pangram—promise to identify AI-generated text, but they're unreliable and sparking a wave of false accusations. The Verge's Emma Roth explains how these systems use AI models to analyze wording, rhythm, and tone, a subjective process that often trips up non-native English speakers. Despite claims of low false positives (Turnitin says under 1%, Pangram 1 in 10,000), real-world cases show significant harm: Minotaur dropped a $2 million book deal with author Jerry Falade over AI suspicions he denies, and a French student sued Yale after a professor used GPTZero to fail him. Even false positive rates of 1% translate to millions of wrongly accused students.

Educators and publishers are adopting these detectors anyway. A Center for Democracy and Technology survey found 43% of US sixth-to-12th grade teachers regularly used AI detectors between 2024 and 2025, and Turnitin's detector auto-enabled in many universities in 2023. The consequences are tangible: a 2023 Stanford study found false flagging of non-native speakers, and an Adelphi student won a lawsuit last February after being wrongly accused. Meanwhile, everyday conversation suffers too—people increasingly accuse each other of "sounding like AI." The article argues these tools replace evidence-based plagiarism checks with unreliable AI judgment, creating an era of distrust where innocent people must prove their humanity.