Security & privacy

iOS flaw let apps record audio after users denied microphone permission

A patched vulnerability in Apple's audio service allowed sandboxed apps to bypass microphone permissions and record without the orange indicator.

Illustration of a broken microphone icon on a phone screen representing a privacy breach.
Illustration generated for this article

Apple has patched a critical privacy flaw in its iOS and iPadOS operating systems that allowed applications to record audio even after users explicitly denied microphone access. The vulnerability, identified as CVE-2026-65339, was discovered independently by security researchers Nik Tsytsarkin of Meta Red Team X and Mustafa Calap, who published their technical analyses recently.

The first fixes arrived in security updates released on August 17, with subsequent patches for older devices following in September and October. Neither Apple nor the researchers have reported any active exploitation of this bug in the wild, but the mechanism reveals a significant lapse in how the operating system verifies process identity during audio capture.

What happened

The core issue allowed a standard, sandboxed application to bypass the Transparency, Consent, and Control (TCC) framework, which is responsible for managing user privacy permissions on Apple devices. In normal operation, when an app requests microphone access, the system prompts the user. If the user selects "Don't Allow," the app should be blocked from accessing the hardware. However, due to this flaw, declining the prompt did not stop the recording.

Nik Tsytsarkin demonstrated that once the initial prompt was declined, the app could still record audio. The prompt would not reappear on subsequent launches, effectively silencing the user's refusal. In his tests, system logs did not show the app using the microphone; instead, they attributed the activity to the audio service itself. This meant that standard auditing tools might miss the unauthorized access.

Mustafa Calap reproduced the issue on iOS 26.6 and an iPadOS 27 beta. His test apps were simple, sandboxed binaries installed via Xcode, meaning they had not passed through Apple’s App Review process. Consequently, it remains unclear if Apple’s review guidelines would have caught such an implementation. Calap noted that in some test builds, the orange microphone indicator in the Dynamic Island or Control Center failed to appear, leaving users completely unaware that their device was listening.

How it works

The vulnerability stemmed from how the audio service, known as audiomxd, validated the identity of the process requesting audio capture. When an app wants to use the microphone, it must prove its identity to the TCC system. This identification relies on an audit token provided by the kernel.

In this flawed implementation, audiomxd accepted the audit token provided by the calling application without sufficient verification. An malicious app could craft a token that claimed to be the audiomxd service itself. Because the service trusted this token, it treated the external app as an internal component. This allowed the app to take a shortcut intended for internal system audio processing, bypassing the TCC permission check entirely.

The kernel inadvertently provided the necessary details for the app to impersonate the service. Once the recording started in the foreground, it could continue even after the app moved to the background or the device was locked, provided the app had the appropriate background mode entitlements. The fix now forces the audio service to verify the caller's true identity directly from the kernel, rather than trusting the token supplied by the client. Additionally, replacing the client identity now requires a private entitlement that third-party apps cannot obtain.

Key details

  • The vulnerability is tracked as CVE-2026-65339 and carries a CVSS score of 5.0 from CISA.
  • Fixes were released in iOS 26.6.1 and iPadOS 26.6.1 on August 17 for newer devices, and in iOS 18.7.11 on October 26 for older hardware like the iPhone XS.
  • The bypass required the app to be already installed on the device and could only initiate recording while the app was in the foreground.
  • System logs attributed the recording to the audio service rather than the app, making detection via standard logging difficult.
  • The orange microphone indicator, a key visual privacy cue, was absent in some test builds during unauthorized recording.
  • Patches were also issued for macOS, tvOS, watchOS, and visionOS, though researchers did not publicly test these platforms.

Why it matters

For software engineers building on iOS, this incident highlights the fragility of trust boundaries within system services. Even with robust sandboxing, a single service that fails to validate its callers can undermine the entire privacy model. It serves as a reminder that permission prompts are only as strong as the backend enforcement mechanisms. Developers must assume that system services may have complex internal logic that can be manipulated if not properly secured against identity spoofing.

For product leaders and security teams, the lack of visible indicators during the exploit is concerning. Users rely on the orange dot and system logs to monitor privacy violations. When both can be suppressed or misattributed, trust in the platform erodes. This flaw demonstrates that privacy features must be defended in depth, ensuring that visual cues and backend logs remain consistent with actual hardware usage.

What you can do

  • Update all iPhones, iPads, and other Apple devices to the latest available OS version immediately to ensure the patch is applied.
  • Review your app’s audio handling code to ensure you are not relying on deprecated or insecure methods for identity verification.
  • Audit your CI/CD pipelines to ensure that any custom entitlements or background modes are strictly necessary and documented.
  • Monitor system logs for anomalies where audio services appear to act without a clear foreground application trigger.
  • Educate users about the importance of the orange microphone indicator and encourage them to report instances where it behaves unexpectedly.
  • If you maintain enterprise apps, verify that your internal distribution channels deliver the latest security patches alongside app updates.
  • #ios security
  • #privacy
  • #vulnerability
  • #mobile development
  • #apple

Tools from the Bytechap store

Keep reading

All stories