TruthLens v1.0.1 - Detection fixes for ChatGPT, blogs, CORS images

🔍 TruthLens v1.0.1 — detection actually works now

Three fixes that turn TruthLens from "loads cleanly but flags nothing" into a working detector on the sites that matter.

🐛 Bug fixes

• Text detector now works on ChatGPT, Claude, Gemini, Perplexity, Poe, Copilot, Substack, Medium, and most blogs. v1.0.0 only scanned <article> and <main p> — missing every <div>-based chat UI and most modern blogs. Selectors now cover the known LLM chat-UI patterns plus a paragraph/div fallback.
• Image pixel analysis no longer silently fails on CORS-tainted canvases. Most third-party hosted images caused the canvas read to throw, leaving pixel scores at zero. We now retry with an anonymous crossOrigin load so the high-freq / entropy / noise heuristics actually run.
• Lowered default minTextLength from 200 → 120 chars. Real-world paragraphs average 80–150 chars; the old default skipped almost everything.

🧹 Manifest cleanup

• Removed unused scripting permission (never used by the service worker) — one fewer permission warning on install.
• Removed empty assets/* from web_accessible_resources.

Highlights (unchanged from v1.0.0)

• 🧠 AI text detector — burstiness, common-word ratio, 25+ LLM phrase fingerprints, syntactic uniformity
• 🖼 Deepfake image detector — known-AI-host check + pixel-level high-freq energy, color entropy, noise uniformity
• 🎙 Cloned voice detector — spectral flatness, silence regularity, pitch jitter on any <audio> or <video>
• 🏷 Inline badges + floating verdict overlay with transparent signal breakdowns
• ⌨ Ctrl+Shift+Y keyboard shortcut and right-click "Scan with TruthLens"
• 🔒 Zero telemetry, no accounts, no servers — audit every line, MIT-licensed
• 🌐 Works on Chrome, Edge, Brave, Opera, Arc, Vivaldi — any Chromium browser

Install in 30 seconds

1. Download truthlens-v1.0.1.zip below and extract it
2. Open chrome://extensions → toggle Developer mode → Load unpacked → pick the folder
3. Pin to toolbar and start scanning

Upgrading from v1.0.0?

Open chrome://extensions and click the reload arrow on TruthLens — or remove and reinstall from the new zip.

Honest limitations

This is heuristic triage, not forensic proof. Every detector on the market has false positives and false negatives — we just say so out loud. The signals are open and documented so you can verify the reasoning yourself.

Links

• 📖 README
• ❓ FAQ
• 🆚 Comparison vs paid detectors
• 📰 Press kit
• 🌐 Landing page

Full changelog: v1.0.0...v1.0.1

If TruthLens helped you spot something, ⭐ star the repo — visibility matters more than money for a tool like this.

TruthLens v1.0.0 - Free AI Detector for Chrome

🔍 TruthLens v1.0.0 — first public release

The free, open-source browser extension that flags ChatGPT-generated text, deepfake images (Midjourney, DALL·E, Stable Diffusion, Flux, Leonardo, Ideogram), and cloned voices — in real time, 100% on-device, with zero telemetry.

Highlights

• 🧠 AI text detector — burstiness, common-word ratio, 25+ LLM phrase fingerprints, syntactic uniformity
• 🖼 Deepfake image detector — known-AI-host check + pixel-level high-freq energy, color entropy, noise uniformity
• 🎙 Cloned voice detector — spectral flatness, silence regularity, pitch jitter (works on <audio> and <video>)
• 🏷 Inline badges + floating verdict overlay with transparent signal breakdowns
• ⌨ Ctrl+Shift+Y keyboard shortcut and right-click "Scan with TruthLens"
• 🔒 Zero telemetry, no accounts, no servers — audit every line, MIT-licensed
• 🌐 Works on Chrome, Edge, Brave, Opera, Arc, Vivaldi — any Chromium browser

Install in 30 seconds

1. Clone or download the source
2. Open chrome://extensions → toggle Developer mode → Load unpacked → pick the folder
3. Pin to toolbar and start scanning

Why this exists

Generative AI now writes, fakes, and clones faster than platforms can moderate. Paid AI-detectors charge $10–$30/month and upload your content to their servers. TruthLens is free, open-source, on-device, and transparent — every verdict shows the underlying signals so you can judge for yourself.

Honest limitations

This is heuristic triage, not forensic proof. Every detector on the market has false positives and false negatives — we just say so out loud. The signals are open and documented so you can verify the reasoning yourself.

Links

• 📖 README
• ❓ FAQ
• 🆚 Comparison vs paid detectors
• 📰 Press kit
• 🌐 Landing page

If TruthLens helped you spot something, ⭐ star the repo — visibility matters more than money for a tool like this.