Privacy policy
Draft of 16 October 2026.
This policy explains how FAME handles your information. FAME is run from Australia. The name of the business behind it isn’t filled in yet. When it says “we”, it means that business.
In short
- We collect what we need to find your tired ad angles and make new ads: your email address, your store’s public pages, the exports and reviews you add, and the ads we make for you.
- AI services do much of the work. Each one gets only what its task needs.
- We don’t sell your data, and the app has no third-party advertising or analytics trackers. One cookie of our own remembers how you first found us.
- You can download a copy of your brand’s data, or delete it, at any time on the Your data page.
What we collect
- Your account. Your email address and the user ID our sign-in service gives you. If you sign in with Google, Google may also share your name and profile picture with that service. Our own database keeps your email address and the ID, when you signed up and when you were last signed in. When you sign up, it also keeps the line about email the sign-in page showed you, which version it was and when, and the terms’ version if the page said signing in agrees to them. When someone asks for a sign-in link, we keep a scrambled form (a keyed hash) of their network’s address for an hour, to limit how many links one network can ask for.
- How you found us, and how far you got. The first time you open our home page, the sign-in page, the waitlist, the store scan, the free angle check or an invite link, a cookie of our own (
fa_ft, below) keeps how you arrived: the campaign tags in the link (such as utm_source), the website that sent you, the page you landed on and when. It never holds your email address. If you then run a store scan, join the beta list or sign up, we keep that with it. We also record when you reach each step, such as signing up, adding a card, adding your store or keeping your first ad, to see where people get stuck. Before you have an account, a step that comes with an email address (asking for a sign-in link, or for a scan’s research) is kept with a scrambled form (a keyed hash) of it, never the address. These records are deleted with your account. - Your store. Its address, and what we read from its public pages: product names, descriptions, types, tags, prices, links and photos, plus your brand kit (colours, fonts, logo, and the words on your homepage, which we use to describe your brand’s voice).
- What you type in. Your brand’s voice, claims to avoid, category and search words, your ad account’s time zone, personas, scenes, brand guidelines, output language, and your edits to our ads.
- Your own winning ads. When you upload one of your brand’s ads to make new versions of it, we keep the image and what we read from it (its message, angle, hook and layout). The image itself isn’t sent to the image service.
- Your team. The email addresses of people you invite, and their role. Everyone on a brand can see the names and email addresses of the others on it.
- Comments. What your team writes on ads and briefs, and who it mentions.
- Feedback. What you send us with Send feedback, which page of the app you sent it from, and whether we may email you about it. It’s emailed to the people running FAME, and deleted with the brand or with your account. Each one you send is counted for an hour, as a scrambled form (a keyed hash) of your email and of your network’s address, to limit how many one person can send.
- Your review of FAME. If you write one when the app asks, we keep what you wrote, the name, role and brand you typed, your rating if you gave one, and the extra ads it got you, if any. We also keep which of the three choices about showing it you picked, the words that choice said, which version they were and when. We show it on our website only if you said we can, and only the way you chose: with your name and brand, or without your name. We never change your words. It’s emailed to the people running FAME, and deleted with the brand or with your account. To have it taken off our website, email hello@famehq.ai. We also keep when the app first asked you, and when you said “Not now”, so it asks at most a few times.
- Creators’ videos. When a creator uses a brief’s upload link, we keep their video, the name and contact details they give, and their note, with the brand. The brand sees them; they’re deleted with the brand.
- Approvals. When a brand sends ads for approval by link, the person who answers types their name. We keep it with each answer, any note they write, the link it came through and the time, with the brand. The brand sees them; they’re deleted with the brand.
- Slack. If you connect Slack, the webhook address you paste, stored encrypted. We post your weekly summary and alerts there.
- Your ad results. The Ads Manager exports you upload: numbers for each ad, such as spend, impressions, clicks and purchases, with the ad, ad set and campaign names. We keep every row as uploaded, so each verdict can show its math. When Claude can sort less than half of an export’s spend into angles, we also keep its answers for that export (the last three such sorts), to find out what went wrong.
- Your Meta ad account, if you connect it. With your permission, we read the same numbers an export has for each ad (spend, impressions, reach, frequency, link clicks and purchases, with the attribution setting, the dates and the ad, ad set, campaign and ad account names), plus each ad’s creative and post IDs, headline, body text, link, start date and video plays, once a day. We also save one copy of each creative’s picture, so your ads can show as cards; only you, your team and anyone you send a share link to can see it. We ask Meta only for read access (its ads_read permission), so we can’t change your ads or budgets. The access Meta gives us is stored encrypted and never shown. Disconnect on the Angles page at any time: we ask Meta to remove our access and delete our stored copy of it, and no more pictures are fetched. Results already pulled, with the pictures saved so far, are kept like uploaded exports. Deleting the brand ends the connection the same way and deletes those too.
- The comments on your ads, if you allow it. When you connect with “Also read the comments on your ads” ticked, we also ask Meta to read your Facebook Pages’ and Instagram accounts’ comments (its pages_show_list, pages_read_engagement, pages_read_user_content, instagram_basic and instagram_manage_comments permissions). Once a day we read the top-level comments on the posts of your ads with the most spend in the last 28 days, and keep each comment’s words, when it was written, its likes and replies, and which post and ads it’s under. We never ask Meta who wrote a comment, so we keep no commenter’s name, ID, username or picture, and we take @handles and email addresses out of the words before we keep them. The words can still include anything else their writer put in them. We use the Page access Meta gives us only while reading and never store it. We never reply to, hide or delete a comment. Comments are deleted 180 days after they were written, when you disconnect, when you take the permission back, if Meta asks us to delete that login’s data, and with the brand.
- Your reviews. Read from your review app’s public widget on your store (Judge.me, Okendo or Yotpo), or from a CSV you upload. We keep each review’s rating, title, text, date and product, and whether it’s verified. We don’t keep reviewers’ names, emails, photos or locations, but a review’s text can include anything its writer put in it.
- The ads we make. Their words, scenes, images and videos, the checks they passed or failed, and whether you kept or rejected them (with your reason, if you gave one).
- Competitors. The brands you follow or hide, and the competitor ads you save.
- Use and costs. Each ad you keep, each image we make and each AI request, with what it cost us. We use this for plan allowances and spending limits.
- Claude connections. If you connect Claude to FAME: the connection’s name, what it’s allowed to do and when it was last used. We keep a scrambled form (a hash) of each access token, never the token itself. If you use the read-only API, we log each call (which tool and when) for a day, to apply rate limits. When an app registers to connect, we keep a keyed hash of its network’s address for a day, to limit how many apps one network can register.
- Payments. If you choose a paid plan, you pay through Stripe and Stripe collects your card details. We never see them. Stripe’s checkout also asks for your billing address, and lets a business add its tax ID and legal name. Stripe keeps these and puts them on your invoices; we don’t copy them. We keep your Stripe customer and subscription IDs, your plan, its status and its billing dates, and a log of the billing events Stripe sends us.
- Free trials. Each email address and each card gets one free trial. When a free trial starts, we keep a scrambled form (a keyed hash) of your email address and, for a trial started with a card, of the code Stripe gives that card (its fingerprint, not the card number). We can’t turn either back into your email address or your card, and they aren’t linked to your account: they only tell us whether an email address or card we’re given has had a free trial before. If it has, the new plan starts without one, and is charged straight away. We keep these after you delete your account, to stop the free trial being taken again.
- Emails. Which emails we sent you and when, and whether the Monday competitor email is on.
- The beta list. Your email address, when we invite you, or when you sign in before you’re invited (that puts you on the waitlist). If you get in with an access code, we keep which code let you in and when. Each code you enter is counted for an hour, as a scrambled form (a keyed hash) of your network’s address and of your email, to limit wrong guesses.
We don’t save the questions you type into Ask. They go to Claude to be answered (see below).
Competitor ads
To show what other brands run, we read Meta’s public Ad Library the way any visitor can, without logging in. We keep each advertiser’s page name, website, Instagram handle and follower count, their ads’ words, dates and links, and reduced copies of the images, at most 800 pixels on the longest side. This library is shared: a page collected for one brand is shown to every brand that follows it. If your page is in it and you want it out, email hello@famehq.ai.
We also keep the words spoken and shown in their video and image ads, as text only. To get them we download a video, take out its sound and a few still frames, and delete the video at once. The sound goes to a speech-to-text service (Groq) and the frames to Anthropic (Claude) to be read, and we delete both as soon as they’re read. We never keep a copy of their videos or sound.
The free angle check
Anyone can run the free angle check at /check without an account. For a check we keep:
- the rows of the two Ads Manager exports you upload (the same ad names and numbers as above) and the results we work out from them, with Claude’s answers when it sorts less than half the spend;
- your ad account’s time zone, and your email address if you add one (it also joins the beta list);
- a scrambled form (a keyed hash) of your network’s address, so we can limit how many checks one network runs. We can’t turn it back into the address, and we delete it after a day.
The check’s pages and the sign-in page use Cloudflare Turnstile to tell people from bots. It runs in your browser and sees your browser details and IP address. We send Cloudflare the result it gave your browser, with your IP address, to confirm it. Cloudflare doesn’t get your files or your email address.
To sort your ads into angles, we send the names, titles and text of up to 200 of your ads (the ones with the most spend) to Anthropic (Claude). Seven days after you upload, the check’s files and results are deleted. We keep only the date and what sorting it cost us. The beta list keeps your email address until you ask us to remove it.
The free store scan
Anyone can scan a Shopify store at /scan without an account. A scan reads the store’s public pages the way any visitor can: its product list, its home page and its About page. Anyone can paste any store’s address, so a scan can be of a store that isn’t theirs. For a scan we keep:
- the store’s address and name, and the facts from its pages we used: the product we picked, that product’s description, and the words on the home and About pages;
- one copy of that product’s photo, shown only on the scan’s own page;
- a scrambled form (a keyed hash) of your network’s address, so we can limit how many scans one network runs. We can’t turn it back into the address, and we delete it after a day.
The scan’s pages use Cloudflare Turnstile as well, in the same way. Cloudflare doesn’t get the store’s address or your email address.
To write the scan’s first angle ideas, we send the store’s name, those pages’ words and the product’s details to Anthropic (Claude). A scan is reached only by its private link, which we never list or share. Seven days after a scan is made, it’s deleted with its photo. If you join the beta list from a scan, the list keeps your email address until you ask us to remove it.
If you ask us to research a scanned store’s ads, you give us your email address, and we look the store up in Meta’s public Ad Library the way we collect competitors (above). We compare its live ads with those of 3 to 5 other brands in its category, and those pages and their ads stay in our shared library. We send the ads’ words and pictures to Anthropic (Claude) to sort them into angles. To make one finished ad, we send the store’s product photo and a description of a scene to fal.ai, then that ad and the photo to Anthropic to check the product looks like the real one. We email you the report’s link, then delete your email address from the scan; it stays on the beta list, as above. The form also says we may send a few follow-up emails about the report, so we keep the words it showed you, their version and when. Each follow-up has a one-click unsubscribe. The report, its ad and its photo are deleted with the scan, 7 days after the scan was made.
How we use it
- To run FAME for you: importing your store and reviews, working out which angles are tired, suggesting new ones, making ads and briefs, and tracking competitors.
- To send you emails: sign-in links, emails about your account (such as your free trial ending, or a warning before an ended trial’s brands are deleted), and emails you can switch off at any time: reminders if your setup isn’t finished, and competitor emails.
- To see where people come from and where they stop, so we can make FAME easier to start.
- To show reviews of FAME on our website, only from people who said we can, and only the way they chose.
- To run plan allowances, payments and spending limits, and to fix problems.
We don’t use your data or your ads to advertise FAME without asking you first.
Who else handles it
These services handle your information for us, under their own terms. We send each one only what its task needs.
| Service | What it does for us | What it gets |
|---|---|---|
| Supabase | Sign-in, and our database | Everything above |
| DigitalOcean | Our servers | The ads we make, your product photos and our error logs |
| Anthropic (Claude) | Sorts your ads into angles, suggests angles, writes ad words and briefs, describes your brand’s voice, drafts personas and scenes, reads the winning ads you upload, finds themes in reviews and in your ads’ comments, checks each image against your product photo, labels competitor ads, and answers Ask | The part of your data each request needs: for example ad names and text, product details, review text, the words of your ads’ comments (never who wrote them), your brand guidelines, a winning ad you upload, your product photo and the image we made. For a store scan, the words on the store’s public pages |
| fal.ai | Makes ad images | Your product photo and a description of the scene. For a store scan’s research, the scanned store’s product photo |
| Google Cloud Vision | Reads small print on product photos, only if we switch it on | Your product photo and the image we made |
| Resend | Sends our emails | Your email address and the email |
| Slack | Gets your weekly summary and alerts, only if you connect it | The message text, which names your angles and their verdicts |
| Stripe | Takes payments | Your email address, and the payment details, billing address and any tax ID you give Stripe |
| Google Fonts | Serves the fonts our pages use | Your IP address and browser details, when a page loads |
| Groq | Turns the sound of competitor video ads into text | The sound of those ads, deleted once read. Never your data |
| Cloudflare (Turnstile) | Checks that a person, not a bot, is using the free angle check or the store scan, or asking for a sign-in link | Your IP address and browser details, on the free check’s and the store scan’s pages and the sign-in page only |
If you connect Claude or ChatGPT to FAME, what you ask about goes to your own account with that app, under your agreement with Anthropic or OpenAI. When we read your store, a review widget or the Ad Library, those sites see requests from our servers. Meta sees the search words we use, but not which brand asked.
Where it’s kept
Our database and servers are in Sydney, Australia. Anthropic, fal.ai, Resend, Cloudflare and Stripe are based in the United States, and these services may handle data in other countries too. Using FAME means your information goes overseas to them.
How long we keep it
- While your account is open, we keep your data so the app can show your history. Apart from the free angle check and the store scan (above), the comments on your ads (180 days after each was written), and a free trial that ends without a paid plan (next), nothing is deleted on a timer yet.
- If your free trial ends and you don’t choose a paid plan, we delete the brands you own, with everything in them and their files, 90 days after the trial ended. We email you 7 days before. Your account stays, so you can still sign in and choose a plan. A paid plan that ends isn’t deleted this way.
- When you delete a brand, its data and files are removed from our database and servers straight away. Copies in our providers’ backups can last until those backups expire.
- Deleting your last brand deletes your account too, unless a paid plan is still running. Cancel it first.
- After a deletion we keep a short record: the date, how many items were removed, and your email address until we’ve finished removing your sign-in record.
- The scrambled forms of the email address and card of a free trial (Free trials, above) stay after your account is deleted, to prevent fraud.
- Stripe keeps its own payment records, and Resend its own records of the emails it sent, under their terms.
Logs
Our servers don’t log the pages you visit. When something goes wrong, we log the error and the page it happened on, so we can fix it. We also keep a short record of each problem (what failed, a trimmed error message with email addresses and links taken out, when, and which brand if any) for 30 days, and email it to ourselves. Errors from background work on your brand are kept with your brand and deleted with it. Our servers’ error logs are kept until they’re rotated out.
Cookies
We use the cookies the app needs to work, and one of our own that remembers how you first found us. There are no advertising cookies, and no one else’s analytics cookies.
| Cookie | What it’s for | How long |
|---|---|---|
fa_at | Keeps you signed in | A day |
fa_rt | Renews your sign-in | 60 days |
fa_pkce | Finishes signing you in | Up to an hour |
fa_login_error | Shows a sign-in problem once | 2 minutes |
fa_oauth_resume | Takes you back to a Claude connection after you sign in | 15 minutes |
fa_meta | Finishes connecting your Meta ad account | 10 minutes |
fa_brand | Remembers which brand you’re looking at | A year |
fa_ft | Remembers how you first found us: the campaign tags in the link, the website that sent you, the page you landed on and when. Never your email address. We use it to see which ways in bring people who stay | 90 days |
fa_agree | Remembers which version of the sign-in page’s lines you saw, so your account keeps it when it’s made | Up to an hour |
Keeping it safe
Pages are served over HTTPS. Sign-in tokens sit in cookies that page scripts can’t read. The ads and videos we make are only shown through links that expire. A brand’s data is only shown to the people on that brand, and to us when we need it to run or support FAME. No system is perfectly secure. If a breach is likely to cause you serious harm, we’ll tell you as soon as we can.
Your choices
- Get a copy. On the Your data page, download your brand’s data as a zip: every row we keep as JSON, plus the files of the ads you kept.
- Correct it. Edit your brand kit, ad words and angles in the app, or email us.
- Delete it. On the Your data page, delete a brand. Deleting your last brand deletes your account too. The app can’t remove your record from our sign-in service by itself, so it records your request and we remove that record by hand.
- Emails. Every competitor email has a link to switch them off, and so does every reminder to finish setting up.
- Questions or complaints. Email hello@famehq.ai. If you’re not happy with our answer, you can complain to the Office of the Australian Information Commissioner at oaic.gov.au.
The Privacy Act
We handle personal information in line with the Australian Privacy Principles in the Privacy Act 1988 (Cth).
Children
FAME is for businesses. It isn’t meant for anyone under 18.
Changes
If we change this policy, we’ll post the new version here with its date. Before a change to how we use your data takes effect, we’ll email you.