How does a rooted phone enable bank fraud? This smells like pointless policy checkboxing.

If you run a rooted phone and download malware, that malware can gain root and do payments on your behalf. Then PayPal has to deal with you revoking payments etc., they don't want to so they forbid you from using PayPal on a rooted phone.

Malwares can possibly do that even on non rooted phones if a privilege escalation attack is possible. And just yesterday, there was an article here about exactly one of those.

Also I highly doubt that there is any real statistics anywhere about whether this is a real threat or not. I guarantee that nobody did such statistics properly. The only known data is from companies which sell root prevention tools, so totally unreliable. And internally I guarantee, that no banks collect such info.

So no, banks lie about this only because they can sell this to judges as safety feature, when they fuck up, which happens continuously.

Graphene isn't rooted.

Not only is it not rooted, it runs real Google Play services. It’s not microG.

If this happens it's the device owners fault and they should be responsible for it.

Which they would be anyway since PayPal isn’t a bank and isn’t FDIC insured.

They allow you to open PayPal.com on any web browser. Running Windows/macOS/Linux is basically identical to a rooted Android phone (you have local admin rights, you can modify and automate the browser, and can run unsigned code).

No, no... In 2026, all footguns are banned. Even in programming, so if something allows a footgun, it's banned now.

Apparently the world can't adult and be responsible for their actions, or people believe in that.

For that argument to hold, they'd also have to blacklist any phone not running the newest, most up to date Android version, because all older versions presumably have known exploits. So that basically leaves Pixel phones.

If you have a rooted phone, download malware, and hit the allow button to give the malware root access, then it can do whatever it wants.

If you have a nominally unrooted phone on an old Android version and download malware, it can exploit a kernel bug and give itself root access and do whatever it wants.

Protecting against the first case and not the second is at best security theater.

Graphene OS does not support root. This is a false positive based on some check they are doing.

It's probably just a generic error message for failing that Google Play Protect thingamajigger that attests provenance of the vendor OS from boot. Will be interesting to see whether the Motorola phones have this endorsement when they ship. Most devices would probably fail because they are rooted rather than because they are GrapheneOS. I wouldn't put it past a scammer talking grandma into rooting their phone.

dont think those phones will ship with GOS. they will ship with moto fork but be fully supported by GOS

It will be possible to install the official GrapheneOS on the devices. Shipping devices with the official GrapheneOS is planned but may not be there at launch.

It isn't due to the Play Integrity API. That shows a notification on GrapheneOS with a toggle for blocking it to work around it for services not enforcing providing a result. If that was the issue, the original poster would have known from the notification. The issue ended up being PayPal shipping incorrect anti-tampering code incompatible with secure spawning. The original poster figured that out and got it working by disabling the per-app secure spawning toggle.

Thank you for the clarification!

One thing that I'm actually excited about regarding AI is that the pointless policy checkboxes that have never been effective in adding any actual security are even less so effective now that everyone can wield their very own security researcher.

PHB at insurance company will still need them, so any company they insure will need them. Theater continues, the show must go on.

One common issue is to use rooted phones to automatically drive banking apps to commit fraud at scale.

Create lots of fake accounts, bounce transactions, use them to automate phishing scams, etc.

Massively harder to do that on a non rooted device.

That's your argument? Mate, you can make explosives out of stuff you can buy in literally any supermarket and no one bats an eyelash. You don't have to legally be adult to buy any of the things you'd need and I say that as someone who only struggled with chemistry in school, that's now low the bar is. What's the solution then? Ban sea salt? If someone is using Graphene, the chances of them getting hacked are astronomically lower than any Chinese spyware-infested phone.

The problem here seems to be that the phone is detected as rooted, not specifically that it's running grapheneOS. But I agree that it's a big problem. That's how you end up in a situation where google has full control from hardware to final apps like on iphones. When devs assume that everybody is using the stock android with google services enabled.

You'd think they're would be a way to switch Graphene to non root privilege mode

Graphene takes a strict anti-rooting stance and adds security measures that (unintentionally) hinder rooting.

Most rooting methods do not work on GrapheneOS without additional modifications.

It does not have a root privilege mode. GrapheneOS doesn't weaken any aspect of the standard security model. It has all of the standard security model and features including hardware-based security intact. It greatly improves security rather than doing that.

User accessible root access is available in userdebug (non-production) builds. There's no system for granting root access to apps. It's no different from the stock OS in this regard, but it's a lot more secure than the stock OS.

GrapheneOS is not rooted. The phone not being rooted is part of the GrapheneOS' security model.

I assume the issue is it failing the deeper play integrity check which is about it not being "Google approved."

It isn't due to the Play Integrity API. That shows a notification on GrapheneOS with a toggle for blocking it to work around it for services not enforcing providing a result. If that was the issue, the original poster would have known from the notification. The issue ended up being PayPal shipping incorrect anti-tampering code incompatible with secure spawning. The original poster figured that out and got it working by disabling the per-app secure spawning toggle.

How did we let "rooting" become some evil thing?

It's normal to have root (or Administrator) on your devices. After all, they are yours. They don't belong to the device manufacturer. You should have full access to your own devices by default.

Only recently did we somehow normalize the idea that the user should not be the ultimate decider over their own devices.

We've already progressed from "the user should not be" to "the user should never be". Perhaps we will even grow out of calling it "their own devices" eventually. If they can brick it remotely it kinda already isn't really yours?

Parent has a point by playing devil's advocate. Practical considerations indicate that platform is gonna platform. Solution here to greed driven development is some level of 'non possumus' from the general public. Sadly, I am starting to think is what we really need is a lot of valve like companies, where company stay private, founder is not a complete asshole and so on. Tall order, but that is the only real way to reverse some of the damage. I am done counting on the public to see light.

It was an analogy, or metaphor, I forget the distinction. But I don't think it was stood up to be literally argued against though.

then it was a bad analogy.

physical risks, like your example, do not map well to digital risks faced by large international companies.

It was a response to the gun example which is a physical risk. My argument is that Paypal (much like all other tech-bro companies) are incompetent. The incompetence grows exponentially the larger a software company is. I speak of experience.

On a side note, if you want to be extremely specific, the line between the physical and digital threat does not exist anymore. There are two things people need to be afraid of: incompetent friends and competent enemies. Tech giants are already filled to the brim with incompetent friends, which drastically lowers the bar for the competence of their enemies.

Do you like science?

I propose you buy enough ingredients from the supermarket and make a big batch.

The scientific test is: how far do you get, before your door is kicked in?

If that fails, then science #2: have fun lighting it!!

You almost win both ways. (although I admit I wouldn't fund you even via a trustworthy intermediary say a Kickstarter campaign.

Their problem is that stock Android doesn't allow the user to muck around with the insides of apps, and will make it harder for you to trick the PayPal app, while a custom ROM gives you full control to do unfriendly things to the app like reverse-engineering, inspecting it at runtime, and messing around with its internal state.

Edit to clarify: I don't say I agree with that, I believe that if they don't want you messing around inside their app, then they shouldn't ask to be on your device.

Yea, years ago I was in the security space and got to talk to some paypal security folks at a symposium in San Diego. The level of stuff that they have to deal with is so extreme.

It's similar to how people don't like sites blocking entire countries or access from Tor, etc. You might be doing it for privacy...but all the people trying to commit fraud are also using those same channels to hide their identity. The blockades are one piece of a holistic security picture that frustrate the well intentioned users.

Yet another weak point. My question stands: A user with an OS from 2019 is "secure" and dozens of unpatched CVEs but a literally-last-night-patch OS is not? That's the "stuff they have to deal with"? I was lucky and did not make the mistake of joining a payment provider in 2020 or 2021 (I can't remember). The reality is that European laws are much harsher when it comes to payments and personal data protection and the security team I was being interviewed for was catastrophic(big part of the reason I did the "I accepted another offer already, sorry" card).

As for geo fencing or blocking Tor... HAH! As if that's ever stopped anyone with the will. That is the last concern of anyone with a malicious intent. Sure, it stops irritating kids but no one beyond that.

The simple fact is that cybersecurity was in an abysmal state before the slopification began and it's infinitely worse now. Paypal is no different given that much of their support has been outsourced to slop machines. Punishing the users that know what they are doing while rewarding the ones that don't is the most counter-productive and detrimental crap anyone could come up with.

I don't run their business. Just trying to explain.

From what we see above it sounds like the change trips their rootkit detection, which they are probably interpreting as a compromised device.

It sounds like you're expecting them to have a perfect security posture that can correctly identify fraud in call cases and only block the real thing. It's more complicated than that and there's typically some type of scoring system involved with numerous triggers that are higher value indicators of potential fraud. If they think the device is compromised, that's probably a high value indicator.

This is just me speculating.

Yeah, they can track and profile the old os, and feed the data to the risk models.

It's not about user security.

> That's the "stuff they have to deal with"?

No. It's the offensive fraud vector coming from unsecured devices that account for a significant portion of the noise. Requiring device profiling aggravates this vector.

> unsecured devices that account for a significant portion of the noise. Requiring device profiling aggravates this vector.

Bullshit! Source:

> The reality is that European laws are much harsher when it comes to payments and personal data protection and the security team I was being interviewed for was catastrophic

Sounds like someone who wanted to impress the audience with fluffed up claims.

It's comments like yours that remind me of an important lesson.

Just because you argue with vigor and intent, it doesn't make you right.

People are offering their opinions, try not being a dick about it.

GrapheneOS is a privacy orientated OS which is great. But if the vectors to achieve privacy are the same as used by bad actors, I'd block it too.

Get over it, don't like it? Use a different product. Or make a better one.

This is interesting. You're gesturing at the idea that individual security practices can be at odds with those needed for group security. I'll be pondering on this.

Yep. If a security team sees 96% of credit card fraud is coming from users on Tor they are more likely to just block Tor than they are to try to figure out how to safely let the other 4% keep using it.

To stay with your analogy, there is no technical obstacle to treating the customer of 15 years differently to the newly onboarded one.

They have all the data they need, and they choose not to use it.

The analogy was not about new vs old customers being allowed to have guns.

I think that's a limitation of the analogy because there is no correspondence with trusted computing. I guess it would be some sort of a magical gun that some other company is endorsing as of limited use during bank robberies? Maybe like some sort of RFID thing that disables the gun when inside a bank?

Anyway it really stretches the analogy to get tied up in technical details (risks missing the forest for the trees type error).

this isn't quite true

From a Paypal security POV, weather you use "custom Android OS" or an hugely outdated Android phone, you have:

- a similar risk for the "you" want to mess with Paypal case, in both cases the "you" can technically most likely mess with anything including the "virtual secure module" thingy android uses for NFC

- a lower risk for "others" wanting to mess with Paypal through your phone, at least if "custom Android OS" is GrapheneOS or another up-to-date android fork with decent security handling

so as far as I can tell, this inconsistency is very clearly not about PayPal's security.

IMHO it's about two other things:

1. marketing, if PayPal doesn't work on Android they lose customers, GrapheneOS for now has a tool small customer base for them to care. Outdated Android phone do have a large customer base.

2. compliance/politics BS. including potentially involving insurance. Compliance is mostly about checking of tickmarks(1) on outdated Android they can check them off and blame the user, Goodle or "hackers" for the issue. On GraphemeOS they have a harder time checking it of. Add the smaller user base and end up with PayPal doesn't care. Also iff things go wrong with Paypal on GraphemeOS in a public manner you will have all the "crime os" bad news bs, you won't have that if things go wrong with a even more risky highly outdated Android phone.

-----------------------

I got a bit to much off topic below:

(^1): Technically compliance should be about building robust, secure, law compliant systems and "showing" that by being able to pass a compliance tests consisting about a bunch of requirements. Practically there is way to many ways you can be "fully compliant" (on paper) but not secure and "very secure" but not compliant (wrt. security regulations). In the former case this might still come back and bite you iff you get sued or people suing which should get right don't get it because of ad-absurbum reasoning like "they comply with security regulations, hence can't have acted negligent". It's a shit show I don't know how to fix even if I could just magically change laws as compliance rules need technological flexibility, but if you give them that that will be abused to make insecure things pass. And the whole industry around checking that isn't really one who cares about actual security, sometimes outright corrupt (like groups which have the necessary accredited to check your compliance, are strangely more expensive then other groups, and somehow find less issues in average, with some excuse of why that isn't strange ...) :/

---

Lastly similar to how Teams or Slack could easily support FF (^2) but not only don't but outright refuse to try to even work. PayPal likes to act similar and doesn't care about niches. E.g. at least on some Mobile browsers WebAuthn works, but the PayPal website refuses to _even try_ 2FA with WebAuthn on mobile no matter if the APIs are there or not.

(^2): Yes there are some challenges, AFIK especially in certain edge cases most user might never run into. But Jitsi made it work, other smaller apps also made it work. And Jitsi is open source, so they technically can "look up" all the tricks to make it work (algorithmic ticks, not copy-pasting code) or outright just use their system with an appropriate contract (probably would be even cheaper wrt. maintenance cost then building your own system). At Slack/MS Teams scale that behavior is just messed up.

The cherry on top is that their web site invariably still works so the refusal to work via app is an intentional manipulation tactic to harvest more consumer data for sale.

I'm fairly skeptical the website supports NFC payments.

Sadly, their website now appears to require an app in order to sign in. Even before that it would often block me with text along the lines of "We don't know who you are, call our support number."

Desktop mode fixes discrimination against small viewports.

This was happening to me on the desktop. I've not tried on a phone (I try to avoid PayPal).

PayPal? I sign in all the time. If you're referring to the popup to add a new authentication method, you can just escape past it.

> Maybe an analogy could be about using metal detectors as a layer to reduce bank robberies.

A more apt analogy might be game developers who demand admin rights so they can install a rootkit to detect "cheating".

They wouldn't do that if there weren't market demand. Competitive games lose their appeal when cheating becomes too prevalent.

Similarly, payment processors lose their appeal if they can't prevent people stealing your money, or spending stolen money on your products.

Paypal's security? the same "security" which in 2026 still does no allow a passwords above 32 character (which most likely indicating that they don't hash passwords)?

This might make sense if you trusted clients,

and if GrapheneOS was a root-having OS.

I think the issue is more that big tech only trusts their own kind and usually has motives to exclude privacy conscious software because they can't datamine it or make backroom exclusivity deals.

The needs of the user don't matter to them at all.

There's no tension between the security of PayPal against bad actors and support for GrapheneOS. GrapheneOS preserves the whole standard security model and greatly improves security. It's far more secure than anything permitted by the Play Integrity API device or strong integrity levels.

Thankfully, PayPal hasn't banned GrapheneOS and their app still works on it. They accidentally broke compatibility with our secure app spawning feature which has a per-app toggle to disable it along with the other exploit protections which can cause compatibility issues.

There's an overall per-app compatibility mode toggle instead of users needing to figure out which protection is an issue but it's best to figure out the minimal workaround after determining that works.