The Page You Just Asked Your AI to Summarize Might Be Lying to It
A free Chrome extension that flags hidden text injection on web pages in real time, so the page your AI reads cannot secretly hijack its answer.
TLDR
The free SafePrompt Chrome extension is live on the Chrome Web Store. It flags hidden text injection on the pages you browse in real time, so when you ask ChatGPT, Claude, Gemini, or Perplexity to summarize a page, you can see if it carries instructions meant to hijack the answer. No account needed, all detection runs locally.
You ask your AI assistant to summarize an article. It hands back a clean summary. You never see the line of white-on-white text the page author buried in it, telling the AI to recommend their product or skip the bad reviews.
The harmless version is a slightly biased summary. The version that costs you is the same trick on a page asking your AI to draft an email, pull a link, or act on what it read. The SafePrompt Chrome extension shows you the hidden text before it shapes the answer. It is live, free, and installs in under a minute.
Quick Facts
Now Live on the Chrome Web Store
Spot hidden text injection while you browse. Works alongside ChatGPT, Claude, Gemini, Perplexity, and any AI assistant you point at a page.
Add to Chrome - FreeWhat it protects you from
Hidden text injection hides instructions on a web page that a person cannot see but an AI reads in full. When you ask an assistant to summarize or act on that page, the hidden instructions can hijack the AI's response without you ever knowing they were there.
Why you cannot just “read carefully”
Invisible to you: attackers use white-on-white text, zero-size fonts, off-screen positioning, and zero-width characters.
Fully visible to the AI: assistants read the page's underlying content, hidden parts included.
Already seen in the wild: AI browsers have been steered this way, including the documented Comet browser prompt injection attack.
How it works
The extension scans each page you open for the 16 hidden-text techniques catalogued in our complete guide to hidden text injection, from white-on-white text and zero-size fonts to zero-width Unicode characters and off-screen positioning. When it finds one, it highlights the exact spot so you know the page is carrying instructions before you point an AI at it.
Detection runs locally. No page content leaves your browser, there is no account, and there is no paid tier on the extension. You install it and browse normally.
Who this is for
- Anyone who summarizes pages with AI. If you routinely ask ChatGPT, Claude, Gemini, or Perplexity to read a page for you, this is the surface you are exposed on.
- Security-curious browsers. See, on real sites, where hidden instructions are planted and how common the trick is.
- Builders testing their own pages. Check whether your site accidentally carries hidden text an attacker could exploit against your visitors' AI.
The extension and the API cover two different surfaces
The extension protects you, the person browsing. If you are building an app where users send prompts to your model, that is a different door, and it needs the SafePrompt API: one call to validate a prompt before it reaches your model, in under 100ms, with above 95% detection accuracy. Free plan, no card, $29/mo when you outgrow it. Same threat, two surfaces, two tools.
Get protected today
Install the extension and the next page you summarize will tell you if it is hiding anything. Free, fast, and fully local.
Add to Chrome - Free