Explore our new Cybersecurity Terms Glossary, explained by our experts.

View the Glossary View the Glossary
PRIVIA

Cybersecurity Terms Glossary

What Is a Cold Boot Attack?

A cold boot attack exploits the fact that DRAM retains its contents for seconds after power is cut, letting someone with physical access recover disk encryption keys from memory.

Illustration for the cybersecurity term Cold Boot Attack
Home / Blog / Cybersecurity Terms / What Is a Cold Boot Attack?
17 November 2026

What Is a Cold Boot Attack?

What Is a Cold Boot Attack?

Full-disk encryption protects data when a device is off. The key, however, has to live in memory while the device is running, and dynamic RAM does not forget instantly when the power stops. A cold boot attack exploits that gap: cut the power, restart the machine from attacker-controlled media within seconds, and read the key straight out of the memory that was never properly cleared.

Why DRAM Remembers

A DRAM cell stores a bit as charge in a capacitor that leaks and must be refreshed thousands of times per second. When refresh stops, the charge decays, but not immediately and not uniformly. At room temperature, useful amounts of data survive for several seconds. Chill the modules and decay slows dramatically: the original research froze chips with inverted compressed air canisters to around minus 50 degrees Celsius and recovered memory contents almost intact after minutes.

J. Alex Halderman and colleagues at Princeton published this as “Lest We Remember” in February 2008, along with tools that scanned a memory image for the key schedules of AES, DES, and RSA. Because an expanded key schedule has strong internal structure, it can be located in a memory dump and reconstructed even when a proportion of bits have already decayed.

The Attack in Practice

The target is a device in a running or suspended state, which in practice means a laptop closed and put in a bag rather than shut down. Two routes exist. The first is to transplant the memory modules into a machine the attacker controls, which works on desktops and servers with socketed DIMMs. The second is to reset the host without cutting power to the modules and boot a small program from USB that dumps memory before the operating system loads.

Vendors addressed the second route by having firmware overwrite memory on boot, a mechanism defined in the Trusted Computing Group specifications. In September 2018, Olle Segerdahl and Pasi Saarinen of F-Secure showed that the flag telling the firmware to perform that overwrite was itself stored in non-volatile memory that could be rewritten with physical access, re-enabling the attack on most laptops then shipping. Several manufacturers issued firmware updates in response.

What Is Actually at Risk

The realistic scenario is a stolen or seized device rather than a remote compromise. Everything resident in memory is exposed: BitLocker, FileVault, and LUKS volume keys, session tokens and browser cookies, cached credentials, private keys held by applications, and anything an open document contained. Where DMA attacks over Thunderbolt are available they are usually faster, but a cold boot attack needs nothing more than a screwdriver and a USB stick.

For a UK organisation, the consequence is a personal data breach assessment rather than only a hardware loss. If a laptop that is lost or stolen can be shown to have been in a suspended state with no pre-boot authentication, the encryption no longer provides the mitigation that the ICO expects to see when judging whether notification is required. Documenting the configuration of your estate is therefore part of the breach response, not an afterthought.

Defences That Hold

None of this is exotic to deploy, and most of it is Group Policy and a build change. Our Hardware Penetration Test Services test laptop builds against cold boot and direct memory access techniques so you know what a stolen device would actually give away.

FAQ – Frequently Asked Questions

Does BitLocker Stop a Cold Boot Attack?

Not in its default configuration. With TPM-only unlocking, the volume key is released automatically during boot and sits in memory whenever the machine is running or suspended, which is exactly what the attack targets. Adding a pre-boot PIN or a startup key means the volume key is never released without the user present, which removes the opportunity.

Is a Cold Boot Attack Still Realistic Today?

It requires physical possession, time, and some skill, so it is not a mass-market technique. It remains realistic for targeted theft, border seizure, and insider scenarios, and the 2018 F-Secure research showed that firmware countermeasures introduced after 2008 could be re-disabled. Memory encryption on recent processors is the change that genuinely narrows it.

What Is the Difference Between Sleep and Hibernate Here?

Sleep keeps the system state in powered DRAM, including the disk encryption key, so the device is a live target the whole time it sits in a bag. Hibernate writes state to the encrypted disk and powers memory down, so a later boot requires the normal unlock path. Setting lid-close and idle timeouts to hibernate is the single most effective policy change.

Can Server Memory Be Attacked This Way?

Yes, and socketed DIMMs make transplanting modules easier than on a laptop. The practical defence in a data centre is physical access control with logging, chassis intrusion detection, and processor-level memory encryption such as Intel TME or AMD SME. For hosted or colocation estates, treat the risk as part of your provider assurance rather than a device setting.

Related Terms

You May Be Interested In These

What Are the CIS Controls?

Cybersecurity Terms

17 November 2026

What Are the CIS Controls?

Read More
What Are Constrained Delegation Attacks?

Cybersecurity Terms

17 November 2026

What Are Constrained Delegation Attacks?

Read More
What Skills Does a Forensics Examiner Need?

Cybersecurity Terms

17 November 2026

What Skills Does a Forensics Examiner Need?

Read More