[hackermatic]

Edit: Other commenters report that Android will silently re-enable cell data under various conditions, so this isn't a surefire solution, either.

The Grugq created a tool for this a decade ago (sadly unmaintained): https://github.com/grugq/portal as part of a presentation about operational security for hackers. It's a great watch if you're interested in how various (in)famous hackers thought they were secure and got busted anyway. https://www.youtube.com/watch?v=9XaYdCdwiWU

[mise_en_place]

> Other commenters report that Android will silently re-enable cell data under various conditions

This is terrifying.

[autoexec]

It's expected. The people who own the phones aren't in control of the OS and the wireless chipsets are closed/proprietary. Cellphones really shouldn't be trusted by anyone.

[mise_en_place]

Correct, the baseband usually has binary blobs. Although I am curious why Google/Apple decided not to make their own baseband, given their new silicon manufacturing expertise.

[adamomada]

IIRC Apple has tried/is trying, but it is ridiculously complex to the point that they had to go back to Qualcomm as there really is no other option. Read: The biggest tech co on the planet stumbles with this, it should be considered a magic box as this point.

[nolist_policy]

Google is sort of trying by using a Samsung modem (instead of Qualcomm) with an IOMMU in between, so at least the modem doesn't have access to the whole address space like on other phones. But they get a lot of flack for it.

[Too]

Armchair speculation: Patents?

[spxneo]

so then whats the other alternative?

solder on some ESPs on an old playstation portable device and connect from starbucks?

[autoexec]

Right now we have no alternatives, but it's not technologically impossible to create mobile devices that give us root access to a mobile OS, or to create open wireless chipsets with open firmware.

[tredre3]

Both Android and iOS will do that when you receive a MMS.

Even if the MMS is supposedly on an intranet, it wouldn't surprise be that a poor implementation might expose the rest of the system to internet for a brief moment.

[nickburns]

i'm almost certain i've had it happen on iOS, too. only reason i can't definitively say—is because i can't rule myself out always having to manually toggle cell data on/off, both radio-level and per-app, when i'm coming/going from my own networks to my mobile VPN.

[spixy]

even in roaming?

[hwbunny]

Just be cautious...