NPC Filter

A filter for likely AI-written posts and comments. Blur, label or hide them on LinkedIn and Reddit. Show any with one click.

Less script. More signal. NPC Filter is a filter for likely AI-written posts and comments. It reads the writing in your LinkedIn feed and tells you when a post looks AI-written, so you can decide what to read. You pick the treatment: - Blur the post until you choose to read it. - Label it as Likely AI or Maybe AI. - Hide it. Every change is reversible. Show any post with one click, and use Why flagged to see which common AI-writing patterns it found. How it works - Rules run inside your browser, for everyone. They look at the writing in posts and comments you scroll past. - Sign in with Google if you want a second opinion from an AI model. When you are signed in, the text of posts and comments the rules did not already flag goes to our server for a model check. Names, handles and profile links are removed first. Signed-out users never send text for a model check. - Free accounts get 500 model checks a month. You stay in control - Pick Blur, Label or Hide, and turn the filter off at any time. - Choose how sensitive it is. Higher settings may flag more human writing. - Press Wrong call? when a flag is wrong. That sends a label, never the text. What it will not do - It does not judge the person. It reads the writing. - It does not promise perfect detection. It gives a best guess, and it can be wrong. - It does not sell your data, show ads or build scores of authors. Right now it works on LinkedIn. Promoted posts and ads are left alone. This is a public beta, so expect rough edges and tell us what you find. Privacy policy: https://npcfilter.com/privacy Questions: privacy@npcfilter.com