LinkedIn AI Detector

Locally detect AI-generated LinkedIn posts using pattern heuristics + Chrome's built-in Gemini Nano. 100% on-device.

Ever scroll LinkedIn and wonder how much of your feed is written by AI? LinkedIn AI Detector scores how likely each post is AI-generated — right in your feed, and entirely on your device. It combines two local signals: fast pattern heuristics for an instant read, and Chrome's built-in Gemini Nano model for an authoritative judgment. Nothing about the posts you read is ever sent to a server. No accounts. No tracking. No cloud. ■ HOW IT WORKS Every post gets a small badge showing an AI-likelihood percentage. A quick heuristic score appears instantly; the on-device model then refines it. Click the ✨ button on any post to (re)run the AI review on demand, and click the score to see the model's plain-English reasoning. ■ KEY FEATURES • Model-first detection — the on-device Gemini Nano model judges posts, catching cleanly-written AI text that keyword filters miss. • Instant badges on every post, colour-coded by likelihood. • One-click ✨ force-review to analyse a specific post with the AI. • Reasoning tooltip — see exactly why a post was flagged. • AI text-watermark detection — flags the invisible characters some AI tools weave into generated text as near-certain (when present). • Auto-hide — collapse high-scoring posts into a slim placeholder you can reveal any time. • Author allowlist — never flag people you trust. • Adjustable sensitivity and per-signal weighting. • Built-in test panel — paste any text to see the heuristic breakdown and the model's score. • Session stats — posts scanned, flagged, and hidden. ■ 100% PRIVATE, 100% LOCAL All analysis runs on your computer using Chrome's built-in AI. The extension makes no network requests and requests only the "storage" permission (to save your settings and cache scores). Your feed is yours. ■ REQUIREMENTS FOR ON-DEVICE AI The Gemini Nano model is optional but recommended. Without it, the extension still works using pattern heuristics. To enable the model you need: • Chrome 138 or newer (desktop) • ~22 GB of free disk space for Chrome to enable the model (the model itself is ~2 GB; Chrome needs the headroom) • A capable GPU (>4 GB VRAM) or CPU (16 GB RAM, 4+ cores) On first install, a welcome page helps you check your device and download the model. You can also manage it any time from the popup or Settings. ■ NOTE ON ACCURACY No AI-detection method is perfect. Treat the score as a strong signal, not proof. The watermark check is near-certain when it fires, but it only catches watermarks that are character-based and that survive copy-paste — so most detection comes from the model. Not affiliated with LinkedIn or Google.