Media
3 items
Description
Scores LinkedIn and Reddit posts for signs of AI-generated writing and shows the score as a badge. A signal, not a verdict. AI Signal puts a small badge above every LinkedIn and Reddit post you read, with a traffic light, a score from 0 to 100 and a mood: 🟢 😌 reads like a person, 🟡 🤨 could be either, 🔴 🤮 reads like boilerplate. Click the badge for the reasons: the stylistic habits it found and how much each one counted. Optionally, hide posts at or above a score you choose. Hidden posts collapse to a one-line stub with a show button, and can be hidden again. The score is a stylistic signal, not a verdict. It looks for habits that machine-written social posts share (the "it's not X, it's Y" hook, em dashes, emoji bullets, stock phrases, uniform sentence lengths, a closing call to action) and habits that people have and models rarely do. It is a first draft: it is better on LinkedIn than on Reddit, it refuses to score very short text, edited or paraphrased text defeats it, and like every detector of its kind it flags some human writers unfairly. Treat red as "reads like boilerplate", not "written by a machine". Everything runs inside your browser. There is no network access, no account, no analytics and no data collection of any kind. Settings are kept in Chrome's extension storage. Open source under the MIT licence: https://github.com/thinkmorestupidless/ai-signal