Skip to content
PRISM

Your data

What happens to what you paste.

Reading Prism needs no account and sets no cookies. Refracting a story needs a Google sign-in. Here is every outside service that sees your link or text, and what Prism keeps and for how long.

Last updated 26 September 2026.

The short version

  • No ads, nothing sold. Reading needs no account and sets no cookies. Refracting a story needs a Google sign-in, which keeps one login cookie in your browser, plus a short-lived one while you sign in. Prism stores no names or emails.
  • Your text leaves Prism. It goes to Google's Gemini AI and Tavily web search, so don't paste anything private or confidential.
  • Google may read it. Prism uses Google's free Gemini plan. Under Google's terms, Google may use what is sent to improve its products, and human reviewers may read it.
  • Kept 12 hours to 30 days. Results are stored under a scrambled fingerprint of what was checked, never under anything that identifies you.
  • Links you refract are saved. Without your name, so the owner can review each one for Prism's Library. Unreviewed ones are deleted after 30 days, and pasted text is never saved. Stories the owner approves are published in the Library, with most of the article's text hidden. Prism also keeps its research on each outlet, never on its authors, for the owner to review for the Library's outlet pages.

Where your text goes

From the moment you press Refract, this is every place your link or text travels, and what each one receives.

You

Paste a link, a YouTube link or text

Prism's server

Reads the page, keeps results for 12 hours to 30 days, and saves links you refract for review

  • Google Gemini AI, backup: Groq

    Gets the text, the claims and the pages found

  • Tavily web search

    Gets search phrases, and the outlet's and author's names

  • rdap.org, Cloudflare, ipwho.is

    Gets the website's address

  • Google fact check archive

    Gets each claim, where available

  • YouTube

    Gets the video link, for its captions

Your browser

Loads the page and shows your results

  • Google icon service

    Gets the websites shown in your results

  • Vercel Web Analytics

    Gets a page visit, without cookies

  • Google Analytics

    Gets a page visit, without cookies

  • Google sign-in, only if you refract

    Gets your sign-in, then your browser keeps a login cookie

The privacy label

Each thing Prism handles, who receives it, why, and how long it stays.

  • The link you paste

    Who gets it
    Prism's server, then the website it points to
    Why
    To download the article, or a YouTube video's captions, and to see which policies the site links to
    Kept
    24 hours as the text read from it, 30 days as the policy links found

  • The article text

    Who gets it
    Google Gemini, or Groq as a backup
    Why
    To sort every sentence and pick the checkable claims
    Kept
    7 days, as the sorted sentences and claims

  • The claims

    Who gets it
    Gemini, Tavily and Google's fact check archive
    Why
    To plan searches, read the sources and reach a verdict
    Kept
    12 hours, per verdict

  • The website's address

    Who gets it
    rdap.org, Cloudflare and ipwho.is
    Why
    To look up its registration and where it is hosted
    Kept
    30 days, per website

  • The outlet's and author's names

    Who gets it
    Tavily, then Gemini
    Why
    To research ownership, funding and reputation
    Kept
    30 days, per outlet and per author

  • The outlet's research

    Who gets it
    Prism's database
    Why
    So the owner can review it for the Library's outlet pages
    Kept
    Until newer research replaces it. Never the author's

  • The title and opening

    Who gets it
    Gemini and Tavily
    Why
    To find other reports of the same event
    Kept
    24 hours, in the Full Picture

  • Your network address

    Who gets it
    Prism's server
    Why
    To stop one person using up the free services
    Kept
    Up to 1 hour

  • Your Google sign-in

    Who gets it
    Google, then your own browser
    Why
    Only if you refract: to check you are signed in
    Kept
    A login cookie in your browser for 30 days, and a 10-minute one while you sign in. Prism's server keeps no name or email

  • How many refractions you've used

    Who gets it
    Prism's server, under a scrambled code
    Why
    To hold everyone to the daily limit
    Kept
    2 days

  • Whether you used the tool while signed in

    Who gets it
    Prism's server, as a count
    Why
    So the owner can see how many people use Prism
    Kept
    Counts that keep no list of who was counted, for 40 days. Nothing links them to you

  • Links you refract, with the article and its results

    Who gets it
    Neon, Prism's database
    Why
    So the owner can review each one for the Library
    Kept
    Up to 30 days while waiting. Approved ones stay until taken down, with the article's text hidden unless its licence allows it. Nothing links them to you

  • Your last result

    Who gets it
    Your own browser tab
    Why
    So reloading the page doesn't lose it
    Kept
    Until you close the tab or press Start fresh

  • Websites in your results

    Who gets it
    Google's icon service, from your browser
    Why
    To show each source's small logo
    Kept
    Not kept by Prism

  • Page visits

    Who gets it
    Vercel Web Analytics
    Why
    To count visits, without cookies
    Kept
    Under Vercel's own terms

  • Page visits

    Who gets it
    Google Analytics
    Why
    Which pages people reach, without cookies
    Kept
    Up to 14 months, by Google

  • Technical logs

    Who gets it
    Vercel, which hosts Prism
    Why
    To find and fix problems. They can include the address you checked and short excerpts of claims
    Kept
    A limited time, by Vercel

Services and their policies

Questions

Prism is a personal project by Prashant Yadav. For questions about your data, email prashantyadavstkld@gmail.com. How each step works is on how it works.