Privacy
This is a site used by children. That single fact decided every choice below, and we would rather write them in plain words than in the usual paragraph of legal fog.
The short version. You can read every story without an account, and signed out we store nothing about you on our servers at all. There is no advertising anywhere on this site, and nothing here profiles a child.
No account, no wall in front of the first story. Which stories you have read, your child's first name if you typed one, their age band and the difficult-stories setting are kept in your own browser, and never sent to us. Clearing your browser data erases them.
Signing in is optional and does exactly three things: it survives a cleared cache, it reaches a second device, and later it will carry a subscription.
We use Google Analytics to count how the site is used, and we have switched off the parts of it that exist to serve advertising. Advertising storage, advertising user data and ad personalisation are all denied by default; Google Signals and ad-personalisation signals are off; IP addresses are anonymised.
Three events are recorded and nothing else: a story was opened, a story was finished, sign-in was started. No child's name, no age, no free text, and nothing that follows anyone across other websites. India's DPDP Act prohibits behavioural advertising to children; the analytics defaults are not on a child's side, so we override them explicitly rather than trusting them.
Two different questions usually get run together here, so we answer them separately.
In making the stories: yes, and here is exactly how. A story's first draft is written with a large language model, working only from a list of claims pulled out of a named source beforehand. Nothing is published from that draft. A person checks the telling line by line against that list, a further adversarial review argues against it from the editions themselves, and a person reads the whole story aloud and approves it before it goes anywhere. Every one of those stages has caught real errors, which is why every one of them exists. Where the tellings genuinely differ, the published page says which one we tell.
While your child is reading: none at all. Nothing is generated at read time. Tonight's story is a fixed, versioned file — the same words tomorrow, the same words offline, and no possibility of a Purāṇa being invented at bedtime.
With anybody's data: none at all. No AI system of ours or anyone else's processes anything about a reader. There is no personalisation model, no recommendation engine trained on behaviour, and nothing about any visitor is sent to a model provider. The three analytics events above are counters. What a child reads stays in their browser unless they sign in, and if they do it sits in their own account and is used to show them their own history — never to train anything, here or elsewhere.
Anyone can report an error from a story page. It takes no account and asks for no name. We keep the story, the version it was reported against, what you wrote, and a two-letter country code — enough to notice a flood of reports, useless for identifying anybody. The form asks you not to include personal details; if you do anyway, that text is deleted once the report has been acted on.
The site is free, needs no account or licence, and can be used by a whole class without anyone signing in — in which case nothing reaches our servers. There is no student roster, no classroom code, and no per-pupil record to administer or delete. If your district needs something in writing beyond this page, write to us and we will answer properly.
Write to sandhyakathas@gmail.com. If you have an account and want it gone, say so and it is deleted along with everything in it — there is no retention period and nothing is kept back.