Graphite study maps AI models' writing tics
What happened
On October 2, TechCrunch reported on a Graphite study of AI writing traits, finding that different models have their own standout word and sentence preferences. The study found that Claude Opus 5.5's clearest AI writing tell is the word "dependable," which appears 23 times more often than in human samples; "this matters" appears 116 times more often. OpenAI's Astra favors "another dimension" and the "not simply X" correction frame, the latter appearing more than 100 times more often than in human samples.
Written by AI from the coverage · updated 1 hour ago
Coverage
Follow the reports to see the story from different sides.
- TechCrunch · AIOpus 5.5 loves to tell you ‘this matters’ (and other AI writing tells)
Graphite 研究发现 Claude Opus 5.5 最显著的 AI 写作破绽是“dependable”一词，出现频率是人类样本的 23 倍，“this matters”更是高出 116 倍；OpenAI 的 Astra 则偏爱“another dimension”和“not simply X”式修正框架，后者出现频率超人类 100 倍。
Heat over time
Not enough continuous observations to draw a trend yet.