If the suspicious text is in a Gmail thread, a LinkedIn post, an X reply, a ChatGPT transcript, or a Substack you are about to forward, the detector has to live in the browser, not in a separate tab. The TextSight Chrome extension puts a one-click detect action on any text you select on the web. Highlight a paragraph, click the toolbar icon, pick a mode in the popup, and the Authenticity Score plus the top flagged sentences come back in two to four seconds. Same multi-model engine, same 0-100 score, same five-band verdict as the web app. The extension is the surface; the backend is identical to what powers app.textsight.ai.
A web detector is fine for one paste. It stops being fine when you are cycling through Gmail, LinkedIn, ChatGPT, and a Substack tab a hundred times a day and dropping the check most of the time.
The friction in a paste-and-scan detector is not the scan itself. It is the tab switch, the manual copy, the visual reorientation to a different app, the paste, the wait, and the trip back. The extension removes those steps. Select a paragraph in a Gmail message, click the toolbar icon, choose a mode, and the score appears in the popup over the page. The flow is the same whether you are reviewing a cold pitch, a LinkedIn comment, or a ChatGPT response you are about to copy into a brief.
Readers who scan suspicious text across an afternoon care less about model quality (every detector is converging there) and more about how cheap each scan is to invoke. The extension makes a single detection a three-second action: highlight, click, mode, read. Over a 50-scan session of cold outreach triage or social feed review, that is meaningful time saved versus bouncing into a separate tab each time. Cheaper scans means more scans, which means catching the things you would otherwise ship.
The extension is not a separate product. It uses your existing TextSight account, the same daily scan quota, the same Authenticity Score logic, and the same detector endpoint as the web app. Sign in once, the extension stays signed in, and your usage from the toolbar counts against the same daily bucket as a paste-flow scan. No second subscription, no extension-only credits, no quota fragmentation between surfaces.
Most standard text surfaces accept select-and-detect. A handful of apps with custom editors fall back to popup paste. Honest list, both columns.
Gmail compose and reading pane. Standard contenteditable and standard message bodies. Highlight a paragraph in an incoming email or a draft, click the extension, read the score. Works on the main view and the reply window.
LinkedIn posts, feed, and direct messages. Both the reading feed and the composer surfaces use a standard editor. The extension scans selected text in posts, articles, comments, and the messaging panel.
Twitter/X timeline, replies, and DMs. Read a viral thread, select a suspicious tweet, scan it before you quote or reply. Short-form text is where the Fast mode does most of its work.
ChatGPT, Claude, and Gemini chat history. Score a model response you are about to paste into a document, or score the prompt you were sent by a teammate. The extension treats the chat transcript as standard selectable text.
Substack, Medium, and most blog readers. Long-form articles where Balanced and Deep modes do their best work. Select two paragraphs, run a scan, read the band before you share the post.
Generic webpage text and contenteditable fields. Comment threads, ticket trackers, CRM notes, news articles, reader-mode pages. If you can select the text, the extension can score it.
Google Docs. The Docs editor uses an iframe-based contenteditable that blocks scripted reads from extensions. The extension drops into popup paste mode: copy your selection, open the popup, paste, scan. The tab switch is still removed, just not the manual copy.
Notion blocks and Notion-style SPAs. Block-based editors with custom keyboard handling do not expose clean selection text to extensions. Same popup paste flow applies.
ATS portals and sanitised editors. Workday, Lever, and similar single-page apps strip formatting and shield input. The popup workflow covers those surfaces honestly without pretending in-place detection works everywhere.
No extension-only tier, no extension-only credits. Your TextSight account quota applies. Yearly billing saves 25% on every paid tier.
Billed $89.88/year — Save $30
Billed $179.88/year — Save $60
Billed $359.88/year — Save $120
Yearly billing saves 25%. View full pricing →
A standard Chrome extension install. About 30 seconds end to end. No build flags, no developer mode, no unpacked-extension warnings.
Open the TextSight listing on the Chrome Web Store and click Add to Chrome. The browser prompts you to confirm the permissions: read selected text on pages where you trigger a detect action, plus storage to cache your sign-in token. Confirm and the extension installs in a few seconds. Same install flow on Edge, Brave, Arc, Vivaldi, and Opera since they all share the Chromium extension API.
Click the puzzle-piece icon in the Chrome toolbar to open the extension menu. Find TextSight in the list and click the pin icon next to it. Pinning keeps the extension icon visible in the toolbar so you can click it without opening the puzzle-piece menu every time. Most users want this on; the extension still works unpinned but the trigger is one extra click each scan.
Click the TextSight icon and sign in with the same email you use on app.textsight.ai. The extension stores an authentication token locally and stays signed in until you sign out manually or the token expires (typically 30 days). One sign-in per browser profile covers every site you use the extension on. Free, Starter, Pro, and Business accounts all carry over automatically; there is no separate extension account.
Open Gmail, LinkedIn, ChatGPT, Substack, an X thread, or any reader-mode article. Highlight the sentence or paragraph you want to score. The extension does not act on its own; it only ever sees text you explicitly select and submit through the popup or the right-click menu. Nothing leaves your device until you click Scan.
Click the toolbar icon (or right-click and choose Detect with TextSight). The popup shows your selected text, the mode toggle (Fast, Balanced, Deep), and a Scan button. Pick a mode, click Scan, and the 0-100 Authenticity Score appears with the five-band verdict and the top flagged sentences. The whole loop is two to four seconds on Balanced.
The mode toggle lives in the popup, not in extension settings. Switching is one click per scan so you can match each piece of reading to the right level of latency and depth.
Single-model classifier optimised for latency. Roughly two seconds for a 5,000-character selection. Right for triage on a social feed, a quick sanity check before quoting a tweet, or a first pass on a long inbox of cold pitches. The five-band verdict is still accurate; the per-sentence breakdown is skipped to keep the response small.
The default and the right mode for most reading. Multi-model ensemble in roughly four seconds. Sentence highlights are included. Use it on incoming emails, LinkedIn posts you are about to share, blog articles you are about to forward, and most everyday detector work. About 80% of extension scans land here.
The most thorough mode. Full per-sentence classifier plus the plagiarism risk indicator on Pro and above. Latency is around eight to twelve seconds for a 5,000-character document. Right when the stakes are real: vetting a freelance writer sample, reviewing a candidate cover letter, checking an investor update before forwarding.
Browser extensions with broad permissions are a real category of risk. This is the honest list of what TextSight requests, what it uses, and what it ignores.
The extension does not read pages passively. It does not scan email bodies, social feeds, chat history, or form fields in the background. The only text that ever leaves your browser is text you explicitly selected and submitted by clicking Scan. Close the popup without clicking Scan and nothing leaves the device. No keylogging, no clipboard monitoring, no DOM polling between scans.
The extension does not inject tracking pixels, third-party analytics scripts, or fingerprinting code into sites you visit. The only outbound network call the extension makes is to the TextSight detector endpoint on api.textsight.ai when you trigger a scan. That call carries the selected text, the chosen mode, and your authentication token; nothing else. No URL logging, no browsing history, no per-site telemetry.
The extension is governed by the privacy policy at www.textsight.ai/privacy.html. Text submitted through the extension is processed for the scan, counted against your daily quota, and not retained beyond the standard logging window applied to web app submissions. Sign-out clears the authentication token and the locally cached preferences (mode default, last-used setting). Manifest V3 architecture keeps the permission surface minimum-grade.
The main detector landing page covering every source model and the full Authenticity Score system.
Open the overview →Sister extension for one-click rewrite on selected text in Gmail, LinkedIn, ChatGPT, and most editors.
See the AI rewriter extension →Android app and iOS web fallback for scanning on a phone, same backend as the extension.
Open mobile guide →Full tier breakdown for Free, Starter, Pro, and Business. Annual billing saves 25%.
See pricing →Same detector, same Authenticity Score, same five-band verdict as the web app. One click on selected text inside Gmail, LinkedIn, ChatGPT, X, Substack, and most pages across the web. Free tier of 3 scans per day works in the extension. Install in 30 seconds.