Considering India's low literacy, having a state owned cyber safety app shouldn't be much of an issue. It's not like a backdoor, but safety of citizens, which is the prime mandate of a sovereign state.
> The November 28 order, seen by Reuters, gives major smartphone companies 90 days to ensure that the government's Sanchar Saathi app is pre-installed on new mobile phones, with a provision that users cannot disable it.
This sounds great in theory. But in practice this sort of thing is rife for abuse. Say, I have complete control over what this app installed on your phone does in the background. And you were my political opponent. Would you trust me to not use this backdoor into your phone to my advantage?
Apps like Netflix, GMail are not forced on users by a govt. It is an open marketplace. Users have options. They are free to buy phones that do not have those apps pre-installed.
This kind of app should be be open source.
> Pre-installed App must be Visible, Functional, and Enabled for users at first setup. Manufacturers must ensure the App is easily accessible during device setup, with no disabling or restriction of its features
While I can get behind the stated goals, the lack of any technical details is frustrating. The spartan privacy policy page[2] lists the following required permissions:
> For Android: Following permission are taken in android device along with purpose:
> - Make & Manage phone calls: To detect mobile numbers in your phone.
> - Send SMS: To complete registration by sending the SMS to DoT on 14422.
> - Call/SMS Logs: To report any Call/SMS in facilities offered by Sanchar Saathi App.
> - Photos & files: To upload the image of Call/SMS while reporting Call/SMS or report lost/stolen mobile handset.
> - Camera: While scanning the barcode of IMEI to check its genuineness.
Only the last two are mentioned as required on iOS. From a newspaper article on the topic[3]:
> Apple, for instance, resisted TRAI’s draft regulations to install a spam-reporting app, after the firm balked at the TRAI app’s permissions requirements, which included access to SMS messages and call logs.
Thinking aloud, might cryptographic schemes exist (zero knowledge proofs) which allow the OS to securely reveal limited and circumscribed attributes to the Govt without the "all or nothing", blanket permissions? To detect that an incoming call is likely from a spam number, a variant of HIBP's k-Anonymity[4] should seemingly suffice. I'm not a cryptographer but hope algorithms exist, or could be created, to cover other legitimate fraud prevent use cases.
It is a common refrain, and a concern I share, that any centralized store of PII data is inherently an attractive target; innumerable breaches should've taught everyone that. After said data loss, (a) there's no cryptographically guaranteed way for victims to know it happened, to avoid taking on the risk of searching through the dark web; (b) they can't know whether some AI has been trained to impersonate them that much better; (c) there's no way to know which database was culpable; and (d) for this reason, there's no practical recourse.
I recently explained my qualms with face id databases[5], for which similar arguments apply.
[1] https://www.pib.gov.in/PressReleasePage.aspx?PRID=2197140&re...
[2] https://sancharsaathi.gov.in/Home/app-privacy-policy.jsp
[3] https://www.thehindu.com/sci-tech/technology/pre-install-san...
[4] https://www.troyhunt.com/understanding-have-i-been-pwneds-us...