The short version
You can use Contxt three ways: at contxt.news, with the Contxt browser extension, or in an AI assistant such as Claude. This policy covers all three. Where something applies to only one, we say so.
- No accounts, no logins, and we don't ask for your name.
- We use what you ask for to write your brief. Your exact search text stays private.
- Briefs are public to anyone with the link, but nothing public connects a brief to you.
- Screenshots are used once and never stored.
- We don't sell your data, run ads, or track you across other websites.
What we collect
- What you ask for. The topic, question, or link you want a brief on. A voice search reaches us as text; your browser converts it, and we never get the audio. A screenshot is read once to find the story, then discarded.
- A random visitor ID (website and extension). Your browser creates it and keeps it, so we can recognize the browser without knowing who you are. It isn't built from your device and resets if you clear site data or reinstall the extension. A second random ID marks each visit and expires after 30 minutes.
- What you do in Contxt. For example, briefs you open and how far you read, searches, shares, and ratings. In the extension, when its button appears and whether you open or dismiss it.
- Technical data. Your IP address and the rough location it suggests (country, state, city), whether you're on mobile or desktop, the page or link that brought you to Contxt, and your timezone, which we use only to know what "today" is.
- Feedback you send. Your ratings and comments, the page you were on, and your email address only if you choose to include one.
We use this to write your briefs, keep Contxt working and free of abuse, and understand how it's used. We don't use fingerprinting or read hardware identifiers.
In AI assistants like Claude
- We receive only what the assistant sends Contxt: the story or link it wants briefed, or which brief to open. The assistant writes this from your request, so it can include details you gave it. We never get your conversation, your account, or images you share with it.
- Requests come from the assistant's servers, so we see their address, not yours. The assistant also tells us its name, version, and which features it supports, plus a random connection ID.
- If the assistant shows Contxt's card, the card loads from contxt.news on your device, which shares your IP address with us, just like visiting the site.
- Your conversation is covered by the assistant's own privacy policy (for Claude, Anthropic's; for ChatGPT, OpenAI's).
In the browser extension
The extension reads a page only when you ask it for a brief on that page. Then it sends us the page's address and, on Reddit, Instagram, and X, a short excerpt of the post. It doesn't log your browsing history, and it talks only to Contxt. Our use of information from the extension follows the Chrome Web Store User Data Policy, including its Limited Use requirements.
Briefs are public
Every brief is saved, and anyone with its link can open it. Links are random and can't be guessed, and we don't list the briefs people request; the homepage shows only briefs Contxt writes itself. Your search text, visitor ID, and IP address are kept private.
A brief's title and question do reflect what was asked, and search engines can find a brief if someone posts its link publicly. Don't ask for anything you wouldn't want visible in a brief.
Services we use
We don't send any of them your visitor ID.
- Render and Cloudflare host the site. Cloudflare's analytics don't use cookies.
- AI and search providers receive what you asked for (a screenshot too, so it can be read), plus public text we gather, to write the brief: OpenAI and Google (through OpenRouter), Brave Search, and Exa.
- ScreenshotOne captures pictures of the public pages and posts a brief cites. X, YouTube, and TikTok tell us a cited post's author and text.
- Langfuse keeps each brief's request and output for a limited time so we can fix problems.
- Formspree delivers the feedback form to us.
On the website, posts and videos a brief cites load from X, TikTok, or YouTube, which can see your IP address and set their own cookies. Exporting a brief as an image loads a code library from Cloudflare. Links out to sources and Google Translate are covered by those sites' policies.
How long we keep it
- IP addresses and city: about 90 days. We keep the country and state.
- Visitor IDs: about 12 months after your last visit, then replaced with random values, so your old activity is no longer tied to you.
- Briefs and search text: kept, so shared links keep working, but disconnected from you on the schedule above.
- Backups and logs: deleted data can remain for up to about two weeks.
The services above keep data under their own policies.
Your choices
Clearing your browser's site data or reinstalling the extension gives you a new visitor ID. To see or delete what we hold, email [email protected] with your visitor ID; since we don't know who you are, it's the only way we can find your records. It stays in your browser until you send it:
We don't change how Contxt works for Do Not Track or Global Privacy Control signals, because we don't sell data or build ad profiles.
Everything else
- Contxt is based in the United States, and our data is stored there. If you use Contxt from another country, your data is processed in the US.
- Contxt isn't meant for children under 13, and we don't knowingly collect their data.
- If this policy changes, we'll update this page and its date.
- Using Contxt is also covered by our Terms of Service.
- Questions: [email protected].