November 19, 2026

Aikido launches agentic pentesting for iOS apps

TL;DR

  • Aikido Security now pentests iOS apps as well, so you can test both your mobile apps and the backend they share in one assessment. Mobile joins the web app and API, external network and AI Capabilities pentests you can already run in Aikido, all with exploit-confirmed findings, AutoFix and retests.
  • Reverse engineering a mobile app got a lot easier this year, with open source agent tooling and virtual jailbroken iPhones sitting on GitHub for anyone to use.
  • iOS apps now face almost the same attack rate as Android apps: 86% versus 89% in 2026, according to Digital.ai.
  • Mobile pentests include a report that can be used for compliance, the fixes and retests all for a fixed price specified upfront. 

Anyone can download your app

Every mobile app you publish is a copy of your code that you hand to strangers. Anyone with an App Store or Google Play account can download it and start pulling apart how it talks to your backend. And they don't wait long. Digital.ai's 2026 Application Security Threat Report, built on telemetry from billions of app instances, recorded first hostile contact within one hour and fifty-six minutes of an app going live in the store. That's probably less time than your release retro.

So it felt like the right moment to finish what we started with Android pentesting. Aikido Security now pentests iOS apps too, which means you can test your Android app, your iOS app and the API they both rely on in a single assessment, with one setup and one report.

Reverse engineering went mainstream this year

Pulling apart a compiled app used to take a specialist and a pile of tools that didn't talk to each other. A wave of open source projects shows how much that has changed in 2026.

Some of them plug reverse engineering straight into coding agents through MCP (Model Context Protocol), giving an agent one consistent way to investigate compiled software. They install and manage the underlying reverse engineering tools themselves, so the person asking doesn’t need to know which decompiler to run. The most popular of these passed 234,000 GitHub stars in about six months. What they give back is pseudocode and a map of how the app’s functions connect. For someone looking for the weak spots in your app, that's usually plenty.

Others boot a virtual iPhone on a Mac using Apple’s own Virtualization.framework. Some ship firmware variants that go all the way up to a fully jailbroken device, with a control socket so an AI agent can drive the virtual phone programmatically. Getting inside an iOS app used to mean hunting down a physical iPhone stuck on an old enough iOS version to jailbreak. Now it's an open source project and a Mac with a few security settings switched off.

Put those two together and you have an agent that can read your app's binary and a jailbroken iPhone to run it on, both free. The skill floor for attacking a mobile app, and an iOS app in particular, has dropped a long way.

iOS stopped being the safe one

For years, iOS was the platform teams worried about less, and the numbers backed that up. In 2023, iOS apps faced roughly half the attack rate of Android apps, a gap Digital.ai says justified much lower security spending on Apple's platform. By 2026 that gap had almost closed: 86% of monitored iOS apps faced attacks, compared to 89% of Android apps (Infosecurity Magazine, reporting on the Digital.ai report).

The report puts this down to AI making attacks cheaper. Attackers use it to reverse engineer apps and generate exploits, and it works across both iOS and Android (SecurityWeek). The extra effort that used to keep iOS apps out of reach doesn't count for much once an agent and a virtual jailbroken iPhone are doing the work.

So if attackers treat your two apps as one target, testing them as two separate projects doesn't make much sense.

Your app is a map of your backend

The most useful thing in a decompiled app is what it reveals about the systems behind it:

  • every API endpoint the app calls, including the ones you never documented
  • how requests are built and authenticated
  • any keys or secrets that ended up bundled into the build
  • the checks you run on the device that never happen on the server

A lot of mobile bugs get exploited through the API, using exactly what the app gave away. A permission check that only lives in the Swift code is a suggestion, as far as anyone calling your API directly is concerned. That's why a mobile pentest that stops at the app misses a big part of the risk.

How Aikido pentests your mobile apps

Aikido's agents install your actual Android and iOS apps, log in as your test users and attack them the way an adversary would, working through the apps and the backend at the same time. Because both apps usually share the same test users and the same API, you set those up once and the findings for both apps and the backend end up in one report.

The difference from an attacker's view is the starting point. Someone reverse engineering your app is working from pseudocode. Aikido starts from your real source code, since mobile pentests are whitebox, and connecting your backend repository as well sharpens the API coverage. Before attacking anything, the agents build a recon report and an attack plan with threat scenarios, so you can follow their reasoning from start to finish.

Every finding comes with an explanation of the risk and reproduction steps your team can follow. On Android those include the exact ADB (Android Debug Bridge) commands to reproduce the issue. Aikido verifies a finding by exploiting it and then stops, so everything that reaches the report has already been proven exploitable.

Assessments run on credits and are priced by scope. Aikido sizes the test to what's actually in your apps and shows you the cost before you launch.

What you need to run a mobile pentest

  • Your mobile source code in a repository connected to Aikido. Mobile pentests are whitebox only for now, blackbox is coming soon.
  • For Android: your APK (Android Package) file.
  • For iOS: your IPA (iOS App Store Package) file.
  • Your backend API URL, plus the backend repository if you want sharper API coverage.
  • Test users with a username and password for each role you want tested.

You can also scope the test to the whole app or point the agents at specific features.

The rest of Aikido pentesting comes along

Everything from the Android launch applies to iOS as well, and it's the same AI pentesting you can already run on your web apps and APIs. You can run AutoFix on a finding and get a pull request with the fix, and mobile pentests produce the same report types as our web pentests, including the audit-ready PDF your compliance team will ask for.

Retests work a little differently on mobile, because the fix has to end up in a new build. Push the fix, rebuild the app, upload the new version from the assessment's overview tab and hit retest. Aikido uses the new build for every retest from then on and marks the finding as solved once it's confirmed fixed.

Your app has probably been unpacked by someone already. Might as well find out what they saw.

Start an mobile Pentest →

What is mobile AI pentesting?

Aikido's agents install your actual Android or iOS app, log in as a user and attack it the way an adversary would, working through the app and its backend at the same time. You get reproducible, exploit-confirmed findings whenever you choose to run it.

Can I test Android and iOS in the same pentest?

Yes. Both apps usually share the same test users and backend, so you set those up once, and the findings for both apps and the API appear in one report.

Do I need two pentests for two apps?

No. Android and iOS can run in one assessment if you want to, priced by scope, and you see the credit cost before you launch.

Do I need to share my source code?

Yes. Mobile pentests are whitebox only for now, so the assessment needs the repository with your mobile app's code. Attaching your backend code is optional and improves API coverage.

Are there requirements for the IPA I upload?

Yes, the IPA needs to be unencrypted and signed.

Are there requirements for the APK I upload?

Yes, the APK needs certificate pinning, root detection and emulator detection turned off so the agents can test the app.

How do retests work for mobile apps?

Upload a new build with the fix and request a retest. Aikido uses that build for every retest after it.

4.7/5
Tired of false positives?

Try Aikido like 100k others.
Start Now
Get a personalized walkthrough

Trusted by 100k+ teams

Book Now
Scan your app for IDORs and real attack paths

Trusted by 100k+ teams

Start Scanning
See how AI pentests your app

Trusted by 100k+ teams

Start Testing
Authors
Mario Popescu
Product Marketing Manager
Subscribe for news
Share:

Recently Added

October 28, 2026

Aikido launches external network pentesting

Aikido launches external network pentesting: AI agents attack your internet-facing systems from the outside and report only findings they actually exploit.
Read more
October 20, 2026

Aikido launches Noon: workforce security for every employee

We built Aikido to secure developers. Today we launch Noon to protect everybody else, from AI-powered phishing and deepfakes to the AI tools security teams can’t see.
Read more
October 6, 2026

Aikido Security achieves FedRAMP Moderate authorization

Aikido has achieved FedRAMP Moderate authorization, bringing AI-native, automated vulnerability remediation to U.S. federal agencies.
Read more