Hacker News Reader

Comment by iancarroll | Hacker News Reader

A bit skeptical of how this article is written as it seems to be mostly written by AI. Out of curiosity, I downloaded the app and it doesn't request location permissions anywhere, despite the claims in the article.

I've noticed Claude Code is happy to decompile APKs for you but isn't very good at doing reachability analysis or figuring out complex control flows. It will treat completely dead code as important as a commonly invoked function.

View on HN

The permissions snippet they show also doesn't include location, and you can't request location at runtime at all without declaring it there.

I'd verify all this stuff for myself, but Play won't install it in my phone so I can't really get the APK. Maybe because I use Graphene...? but I don't know all the ways they can restrict it, maybe it's something else (though for a pixel 9a it's rather strange if it's hardware based).

--- EDIT ---

To be specific / add what I can check, this is what my Play Store "about -> permissions" is showing:

    Version 47.0.1 may request access to
    Other: 
      run at startup
      Google Play license check
      view network connections
      prevent phone from sleeping
      show notifications
      com.google.android.c2dm.permission.RECEIVE
      control vibration
      have full network access

which appears fairly normal, and does not include location, and I think Play includes runtime location requests there. Maybe there's a version-rollout happening, or device-type targeting?

If you use Aurora Store instead of the Play store, you can download APKs. They are a Google Play store proxy.

Or from APKPure https://apkpure.com/the-white-house/gov.whitehouse.app

Is there a way to conver that xapk format to apk other than installing their app?

yes, unzip it.

I have many apps that refuse to work. They try to open play store app which does not have logged in account.

The app doesn't work

The aurora store will identify whether apps require google play services before you try to install them.

> it doesn't request location permissions anywhere, despite the claims in the article

The article does not claim the app requests the location. It claims it can do it with a single JS call.

> The article does not claim the app requests the location. It claims it can do it with a single JS call.

so can ... any other code anywhere on a mobile device? That is how API work...

You need to state the permissions you *may* request/use in AndroidManifest.xml. This data can then be displayed to users pre-installation.

From the (limited) article, it doesn't seem they do this: https://thereallo.dev/blog/decompiling-the-white-house-app#p...

----

EDIT: I'm mistaken. From the Play Store[0] it has access to

* approximate location (network-based)

* precise location (GPS and network-based)

[0] https://play.google.com/store/apps/details?id=gov.whitehouse...

This seems to disagree with:

> The location permissions aren't declared in the AndroidManifest but requested at runtime

*shrug*, someone should dig deeper. It looks like the article may not match reality.

What version do you see? 47.0.1 doesn't have that for me: https://news.ycombinator.com/item?id=47557033

Very unusual: 47.0.1 is showing these permissions when on my MacBook viewing the store entry.

The Play Store doesn't show these permissions when viewed on my Pixel 9 Pro, and the APK doesn't have these permissions when downloaded/extracted.

It can request with a JS call. It can't passively collect it without you approving first. The article is written like calling that JS function will turn on location tracking without consent.

He explicitly says he can't determine it, but that the location tracking as configured will turn on once the user grants consent. All true statements.

How would you have written it differently

"If the user chooses to opt-in and grants location-tracking permission, the app is then, and only then, able to track the user's location?"

But that's not true; it could easily fallback to other forms of geolocation like using the current IP.

Good lord. So could literally any app on the planet

That would allow you to see the local network IP (not actually sure you even get that, tbh). To get more detailed information about IP configuration, you need Location permission. Been there, done that. Most Android network information calls provide degraded information if you have not been granted Location permissions.

If an app can make an HTTP request, the app can know the user's public IP address and the geolocation derived from that.

This data has well-known limitations, but I think it is the fallback people are talking about here.

You would be lying if you wrote that because you do not know if that is true.

>as it seems to be mostly written by AI.

Is there something in particular that made you conclude that or are you going just with how it felt?

For what it's worth, it didn't seem to me.

Perhaps the apparent hallucination they mentioned in their comment?

Apparently just like OP, you didn't read the article either. Just because the app doesn't ask for permission in the manifest doesn't mean it can't be acquired at runtime. It's very publicly documented [0].

So, no. Not a "hallucination".

[0] https://documentation.onesignal.com/docs/en/location-opt-in-...

How certain are you of that?

That appears to be about providing a message to the user before requesting permissions.

However, it appears even permissions you allow your app to request still need to be declared beforehand? https://developer.android.com/training/permissions/requestin...

Regardless, people are reporting mixed info on whether the app declares location access: https://news.ycombinator.com/item?id=47557010

I checked all versions. Maybe, just maybe, the app was changed in response? Hmmm, I wonder...

https://imgur.com/a/SNJL4XO

This is incorrect. On Android, you must do BOTH to actually get location APIs to work.

Well, I will argue that you are incorrect and do one better and ask why a Huawei SDK [0] is embedded in the app beyond the location tracking?

[0] https://www.sambent.com/the-white-house-app-has-huawei-spywa...

You mean fabrication?

There's a specific writing style for globalized English that AI's use. And then this post also had none of the stylistic flourishes that a real author might add. And then simple things like constructing a table of 68 libraries or whatever organized by relatively subjective categories. That is something that nobody is going to do by hand.

There is a new term "load-bearing" which is used a lot in my usage of AI. Has anyone else encountered this term being used a lot in their conversations? Or is it a quirk of personalization?

been getting a lot of "load-bearing" and "roll your own" lately.

us humans, even if kinda trash at many things, are pretty rad at pattern recognition.

I use load-bearing all the time in conversation. People need to be careful that just because they don’t use certain phrases, it doesn’t automatically mean AI.

I use it all the time, but almost always sarcastically (as in "load-bearing tinyproxy instance").

just what an AI bot would say! ;)

I've heard it a lot from podcasts that are towards the abundance movement. I think its common within the rationalise movement.

Personally I really like it for "load-bearing assumptions". Because it let's you work with assumptions whilst pointing out the potential issues of that assumption.

There are also fashions. So people could be using "load-bearing" more because it's fashionable. Like "lets double-click on that", or "spinning rust", etc

Both you and parent are making a lot of load-bearing assumptions.

As someone who likes to use a lot of em dashes in writing -- the 'heuristics' that AI 'hunters' like to use need a lot of further refinement before I would trust them with anything. And yet there are legions of anti-AI crusaders out there wielding them like weapons.

These folks are reinforcing a bias against all kinds of people, particularly those who are not native English speakers and were very likely taught 'globalized' English in their language training.

what version are you on?

from the iphone app store: version 47.0.1 - minor bug fixes - 34 minutes ago

while the parent posted 18 minutes ago

they may have patched the location stuff as part of the “minor bug fixes”?

I have the iOS version from yesterday, haven't updated the app yet.

No location permission request prompting encountered. In system settings, where each app requesting location data is listed, it isn't present either.

Ah. So another way to say it doesn’t get your location every 4 seconds.

Isn’t it useless to talk about the iOS version if the article is about the Android app?

how do you know it didn't lie during the decompilation?

It doesn't have to lie: unfortunately libraries that are essentially a full application themselves (complete with their own permissions) are not uncommon on mobile.

So it could come across a manifest that includes location permissions and some code that would (if enabled) send location, but it might do a bad job properly tracing

I think you should make proper counter arguments instead of dismissing something because they used a specific tool.

Ad-HomineLLM is a logical fallacy IMO and adds little value. I would hope eventually HN and other sites add this to the guidelines similar to other claims like vote manipulation etc.

Sorry, making up a word to try and frame distrust of LLM-generated content as a “logical fallacy” is a bad take.

HN doesn’t have guidelines against anti-LLM rhetoric, but it does for LLM-generated comments.

> Don't post generated comments or AI-edited comments. HN is for conversation between humans.

https://news.ycombinator.com/newsguidelines.html#generated

GP was arguing against the OP, not a comment, and AI written posts are fair game.

Also, the comment you responded to was criticizing the attack to the substance of the post based on who/what wrote is. The comment neologism actually fits, IMO.

Reply on news.ycombinator.com