A significant vulnerability in Apple’s T2 security chip has been highlighted by @checkra1n and it has serious implications as to how device security is approached in the years to come.

Here’s why...

(a #CyberSecurity thread) 1/

Firstly, what is the T2 security chip?

It is Apple Silicon’s bespoke System on a Chip (SoC) responsible for device security. It manages encryption utilities, camera/audio controls and is heavily integrated with security features such as Touch ID.

It’s pretty important. 2/

First launched with the iMac Pro 2017, it runs a separate operating system known as BridgeOS (which is strangely a derivative of WatchOS).

The SoC is now embedded into nearly every major Apple device, including the iMac, iMac Pro, Mac Pro, MacBook Air and MacBook Pro. 3/

The T2 itself is a 64-bit chip with an ARMv8 architecture, containing a Small Enclave Processor (SEP) which is modelled closely on that of Apple’s A10 Fusion chip (essentially the general-purpose processor used in the iPhone 7). 4/

In 2019, the ‘Checkm8’ exploit was published by @checkra1n with the A5-A11 SoCs being vulnerable.

This allowed for the semi-tethered jailbreaking of the affected iOS devices, effectively providing root access to them. 5/

This exploit was particularly significant because it targeted a vulnerability within the Boot read-only-memory (BootROM), meaning there is absolutely no way for Apple to patch the issue through iOS, unlike most other jailbreaks.

The A12-A14 SoCs do not appear vulnerable. 6/

In early October of 2020, a new @checkra1n release, primarily developed by @axi0mX, provided support for the Checkm8 exploit to be used to boot arbitrary code on the T2 SoC through a Device Firmware Upgrade (DFU) reboot. 7/

As the exploit is semi-tethered, physical access to the device is required for it to be used (typically via USB).

If deployed successfully, root access can be gained on the T2 chip enabling control of most of its functionality. 8/

How can this be used maliciously?

As far as we know currently, the risk is reasonably low for devices containing the T2 SoC.

This is despite a lot of sensationalist misinformation claiming otherwise. 9/

It does not provide immediate access to encrypted disks without knowledge of the device’s password and as far as I can tell, there is no persistence of the booted code.

This means it is unlikely for now that it can be used effectively with other exploits such as a keylogger. 10/

If the security threat is low, then why is this significant?

Well the T2 SoC has commonly been described as a black box for Apple’s device security mechanisms, and it has now been seriously compromised. 11/

The most critical part of this is that it cannot be patched.

Many key security barriers have been bypassed which is still concerning despite a negligible present threat, and yet Apple are unable to do anything about it. 12/

The obvious question now is how did Apple find themselves in this position?

Using a dedicated trusted SoC for securing certain services/processes has been in ways a long-term experiment by not just Apple, but other Tech manufacturers such as Intel, Cisco and Samsung. 13/

The argument is predicated upon the idea that these security black boxes are impenetrable, and can therefore be trusted with ensuring device security.

The fact that the T2 chip has been penetrated, albeit without much risk, fundamentally undermines this argument. 14/

Flaws in such SoCs are cemented into a device’s hardware, meaning that if they are discovered little can be done to correct them.

This vulnerability in the Apple T2 chip has highlighted the risk that these systems carry. 15/

If the exploit could be used more maliciously, we could be dealing with a seismic security crisis of which little could be done to correct.

This has proven that even the most rigorous engineering can be flawed and that no dedicated SoC will be completely secure. 16/

This therefore raises the question, is the use of dedicated SoCs for managing a device’s security worth the risk?

Should this approach be continued or is a rethink necessary?

I think we can all agree that at the very least this conversation needs to be had. 17/