AMD’s bulletin identifies CVE-2026-43613 and lists target months for planned mitigations. Separately, researchers reported two physical lab demonstrations that recovered test PINs by rolling back TPM failed-attempt state. The results came from specific laptop configurations, and Intel says its test did not demonstrate a vulnerability in an Intel component.
What researchers demonstrated
Researchers reported recovering a four-digit Windows Hello test PIN in 236 attempts over 57 minutes on an HP OmniBook 3 with an AMD Ryzen AI 5 330. The Windows setup required a restart after every five incorrect entries, but the demonstrated rollback prevented the TPM lockout delay from applying.
On a Framework Laptop 13 with an Intel Core Ultra 5 125H and Winbond flash, researchers recovered a six-digit BitLocker test PIN in 236 attempts over 6 minutes and 55 seconds.
| Protection tested | Laptop and processor | Test PIN | Attempts | Demonstration time |
| Windows Hello | HP OmniBook 3; AMD Ryzen AI 5 330 | Four digits | 236 | 57 minutes |
| BitLocker | Framework Laptop 13; Intel Core Ultra 5 125H; Winbond flash | Six digits | 236 | 6 minutes 55 seconds |
The paper, Hello Again, describes the researchers’ work on replay-protected monotonic counters (RPMC), mechanisms intended to prevent older security state from being restored. Version 1 was submitted to arXiv on November 13, 2026. The authors are Christian Werling, Matti Finder, Hans Niklas Jacob, Julian Prein, Niclas Kühnapfel, and Jean-Pierre Seifert, affiliated with Technische Universität Berlin (the paper record).
The processor matters, but so does the flash implementation and the physical access required by these demonstrations.
How TPM counter rollback affects the lockout
A Trusted Platform Module (TPM) keeps track of failed PIN attempts so repeated guesses trigger a lockout. In the reported attacks, researchers targeted that stored state, rather than breaking the PIN encryption itself. The state is persisted on SPI flash, and RPMC is meant to protect its counter from rollback.
The paper describes two distinct weaknesses. In one, an AMD nonvolatile controller fails to verify that an RPMC increment succeeded, allowing an interposer to suppress the increment while reporting success. In the other, voltage glitching can expose a key used by Winbond flash to authenticate counter values, enabling counter forgery. These findings concern specific implementations, not every computer using an AMD or Intel processor (the research paper).
Intel’s position on the Framework Laptop test
Intel says the research did not demonstrate a vulnerability in an Intel component. Its November 16, 2026, notice describes the reported attack as requiring physical access, specialized equipment, and a vulnerable third-party SPI-flash implementation. The Framework test used Winbond flash, and Intel recommends that customers ask their system manufacturer whether updates are available (Intel’s security announcement).
AMD’s planned mitigation targets
AMD’s bulletin identifies CVE-2026-43613 as an RPMC increment-suppression vulnerability. It lists target months by processor family:
| AMD processor family or product row | AMD target month |
| Ryzen 3000 desktop, Ryzen 4000 desktop, Ryzen 5000 desktop, Ryzen 7020, Ryzen 7045, Ryzen 9000 desktop, Ryzen AI Max 400 | November 2026 |
| Ryzen 4000 and 5000 mobile, Ryzen 6000, Ryzen 7000 desktop, Ryzen 7030/7035/7040/8040, Ryzen 8000 desktop, Ryzen AI 300/400, Ryzen AI Max 300, Threadripper 7000/9000, and several Ryzen Z-series entries | December 2026 |
Some processor families have multiple firmware branches and target months. AMD’s bulletin presents these dates as planned target releases (AMD security bulletin).