5 ms·

You're framing this as if it's all about the bounty and Apple just hasn't gotten around to it yet. That's only a small fraction of the story and could easily be

You're framing this as if it's all about the bounty and Apple just hasn't gotten around to it yet. That's only a small fraction of the story and could easily be forgiven. If I wanted to make Apple look good, I'd focus on that part, but that would be rather biased to ignore the whole picture... Tokarev discovered 4 iOS 0-days, then reported them all to Apple back in May. After months of Apple's continued refusal to fix or even publicly acknowledge all four of the issues, Tokarev made all of them public on GitHub. Weeks passed, and now it's today. Apple has yet to fix or publicly acknowledge two of the four security vulnerabilities. That should make Apple look bad because it's some fundamentally irresponsible security practices. Yes, I'm biased. I'm human, not a computer, and it's stuff like this that makes me biased towards Apple. They should receive negative publicity for this, then they should change how they do things. At the very least, app developers and users should be warned about the two issues that have yet to be fixed.

tptacek 5y ago
I don't know anything about Apple's bug bounty program except that there's a prevailing attitude that it is not the well-oiled machine that Google's bug bounty program is perceived to be, and I'm not super interested in making a case for Apple here. But because this is a recurring theme in every discussion about every bug bounty run by anyone: * There are valid reasons that bugs can take longer to fix than you'd expect; the most notable of them is when the bug you found is actually systemic, or has a deep root cause, and the real fix for the vulnerability is more complicated than the surface bug. Without a hard timeline, some shops will work to get the root cause fixed on some bugs even at the cost of an increased timeline, because the patch for the surface bug reveals the pattern and amplifies risk to customers. * As a reporter, you can take some measure of control over the process back by providing a fixed timeline (like the P0 90 days). There's no negotiation needed; you give the vendor time to fix and they either do or don't, but either way you're going public. That is a valid way to go about things, but may cost you the bounty. * These things are bug-dependent, and the process that runs for a zero-interaction RCE won't be the same as the process that runs for a bug that requires a malicious app store app and only gives access to the contact database. * Message boards tend to expect that big vendors can just shell out for the bounty as a show of good faith. It's easy to see why they believe that. It makes sense. But it also creates broken incentives. The limiting reagent on bugs isn't bounty dollars (these are indeed barely even rounding errors to major vendors), but rather programmer time. If you pay out for weak, stuck-in-process bugs, you create incentives that redirect programmer time to those weak bugs and away from more significant bugs; as angry as you can reasonably be about a malicious app being able to snarf your contacts, if you're rational, you're a lot more concerned about memory corruption flaws, which is what you really want people spending their time on.
- nopcode 5y ago
  > * These things are bug-dependent, and the process that runs for a zero-interaction RCE won't be the same as the process that runs for a bug that requires a malicious app store app and only gives access to the contact database. It would be interesting to understand at what point this becomes a GDPR issue, and if the GDPR legislation can be used to pressure companies in expediting this process.
  - tptacek 5y ago
    There's a "be careful what you wish for" argument here, because my understanding is that the FAANG vendors are snapping up security people just as fast as they possibly can, and, again, ceteris paribus you'd rather have those people working on the actual most serious vulnerabilities rather than the ones causing the noisiest bounty drama. But you could reasonably go either way on this I guess.
    - jacquesm 5y ago
      The degree of damage done by particular vulnerabilities is different from person to person. For one individual losing their contact database to some un-identified third party might be a 'meh' event, for another it could be a disaster.
- saagarjha 5y ago
  > If you pay out for weak, stuck-in-process bugs, you create incentives that redirect programmer time to those weak bugs and away from more significant bugs; as angry as you can reasonably be about a malicious app being able to snarf your contacts, if you're rational, you're a lot more concerned about memory corruption flaws, which is what you really want people spending their time on. I don't understand how this isn't already reflected in bug bounty pricing tiers? Like, if I access your contacts, I get $x, but if I can access all your photos, I get $xx. As a user, I couldn't care less for how my data got accessed, whether it's a logic bug or memory corruption…
  - tptacek 5y ago

tptacek 5y ago

I don't know anything about Apple's bug bounty program except that there's a prevailing attitude that it is not the well-oiled machine that Google's bug bounty program is perceived to be, and I'm not super interested in making a case for Apple here. But because this is a recurring theme in every discussion about every bug bounty run by anyone: * There are valid reasons that bugs can take longer to fix than you'd expect; the most notable of them is when the bug you found is actually systemic, or has a deep root cause, and the real fix for the vulnerability is more complicated than the surface bug. Without a hard timeline, some shops will work to get the root cause fixed on some bugs even at the cost of an increased timeline, because the patch for the surface bug reveals the pattern and amplifies risk to customers. * As a reporter, you can take some measure of control over the process back by providing a fixed timeline (like the P0 90 days). There's no negotiation needed; you give the vendor time to fix and they either do or don't, but either way you're going public. That is a valid way to go about things, but may cost you the bounty. * These things are bug-dependent, and the process that runs for a zero-interaction RCE won't be the same as the process that runs for a bug that requires a malicious app store app and only gives access to the contact database. * Message boards tend to expect that big vendors can just shell out for the bounty as a show of good faith. It's easy to see why they believe that. It makes sense. But it also creates broken incentives. The limiting reagent on bugs isn't bounty dollars (these are indeed barely even rounding errors to major vendors), but rather programmer time. If you pay out for weak, stuck-in-process bugs, you create incentives that redirect programmer time to those weak bugs and away from more significant bugs; as angry as you can reasonably be about a malicious app being able to snarf your contacts, if you're rational, you're a lot more concerned about memory corruption flaws, which is what you really want people spending their time on.

nopcode 5y ago
> * These things are bug-dependent, and the process that runs for a zero-interaction RCE won't be the same as the process that runs for a bug that requires a malicious app store app and only gives access to the contact database. It would be interesting to understand at what point this becomes a GDPR issue, and if the GDPR legislation can be used to pressure companies in expediting this process.
- tptacek 5y ago
  There's a "be careful what you wish for" argument here, because my understanding is that the FAANG vendors are snapping up security people just as fast as they possibly can, and, again, ceteris paribus you'd rather have those people working on the actual most serious vulnerabilities rather than the ones causing the noisiest bounty drama. But you could reasonably go either way on this I guess.
  - jacquesm 5y ago
    The degree of damage done by particular vulnerabilities is different from person to person. For one individual losing their contact database to some un-identified third party might be a 'meh' event, for another it could be a disaster.
saagarjha 5y ago
> If you pay out for weak, stuck-in-process bugs, you create incentives that redirect programmer time to those weak bugs and away from more significant bugs; as angry as you can reasonably be about a malicious app being able to snarf your contacts, if you're rational, you're a lot more concerned about memory corruption flaws, which is what you really want people spending their time on. I don't understand how this isn't already reflected in bug bounty pricing tiers? Like, if I access your contacts, I get $x, but if I can access all your photos, I get $xx. As a user, I couldn't care less for how my data got accessed, whether it's a logic bug or memory corruption…
- tptacek 5y ago