Entries Tagged "physical security"

Page 1 of 25

New Attacks Against Secure Enclaves

Encryption can protect data at rest and data in transit, but does nothing for data in use. What we have are secure enclaves. I’ve written about this before:

Almost all cloud services have to perform some computation on our data. Even the simplest storage provider has code to copy bytes from an internal storage system and deliver them to the user. End-to-end encryption is sufficient in such a narrow context. But often we want our cloud providers to be able to perform computation on our raw data: search, analysis, AI model training or fine-tuning, and more. Without expensive, esoteric techniques, such as secure multiparty computation protocols or homomorphic encryption techniques that can perform calculations on encrypted data, cloud servers require access to the unencrypted data to do anything useful.

Fortunately, the last few years have seen the advent of general-purpose, hardware-enabled secure computation. This is powered by special functionality on processors known as trusted execution environments (TEEs) or secure enclaves. TEEs decouple who runs the chip (a cloud provider, such as Microsoft Azure) from who secures the chip (a processor vendor, such as Intel) and from who controls the data being used in the computation (the customer or user). A TEE can keep the cloud provider from seeing what is being computed. The results of a computation are sent via a secure tunnel out of the enclave or encrypted and stored. A TEE can also generate a signed attestation that it actually ran the code that the customer wanted to run.

Secure enclaves are critical in our modern cloud-based computing architectures. And, of course, they have vulnerabilities:

The most recent attack, released Tuesday, is known as TEE.fail. It defeats the latest TEE protections from all three chipmakers. The low-cost, low-complexity attack works by placing a small piece of hardware between a single physical memory chip and the motherboard slot it plugs into. It also requires the attacker to compromise the operating system kernel. Once this three-minute attack is completed, Confidential Compute, SEV-SNP, and TDX/SDX can no longer be trusted. Unlike the Battering RAM and Wiretap attacks from last month—which worked only against CPUs using DDR4 memory—TEE.fail works against DDR5, allowing them to work against the latest TEEs.

Yes, these attacks require physical access. But that’s exactly the threat model secure enclaves are supposed to secure against.

Posted on November 10, 2025 at 7:04 AM • View Comments

Regulating AI Behavior with a Hypervisor

Interesting research: “Guillotine: Hypervisors for Isolating Malicious AIs.”

Abstract:As AI models become more embedded in critical sectors like finance, healthcare, and the military, their inscrutable behavior poses ever-greater risks to society. To mitigate this risk, we propose Guillotine, a hypervisor architecture for sandboxing powerful AI models—models that, by accident or malice, can generate existential threats to humanity. Although Guillotine borrows some well-known virtualization techniques, Guillotine must also introduce fundamentally new isolation mechanisms to handle the unique threat model posed by existential-risk AIs. For example, a rogue AI may try to introspect upon hypervisor software or the underlying hardware substrate to enable later subversion of that control plane; thus, a Guillotine hypervisor requires careful co-design of the hypervisor software and the CPUs, RAM, NIC, and storage devices that support the hypervisor software, to thwart side channel leakage and more generally eliminate mechanisms for AI to exploit reflection-based vulnerabilities. Beyond such isolation at the software, network, and microarchitectural layers, a Guillotine hypervisor must also provide physical fail-safes more commonly associated with nuclear power plants, avionic platforms, and other types of mission critical systems. Physical fail-safes, e.g., involving electromechanical disconnection of network cables, or the flooding of a datacenter which holds a rogue AI, provide defense in depth if software, network, and microarchitectural isolation is compromised and a rogue AI must be temporarily shut down or permanently destroyed.

The basic idea is that many of the AI safety policies proposed by the AI community lack robust technical enforcement mechanisms. The worry is that, as models get smarter, they will be able to avoid those safety policies. The paper proposes a set technical enforcement mechanisms that could work against these malicious AIs.

Posted on April 23, 2025 at 12:02 PM • View Comments

Criminal Gang Physically Assaulting People for Their Cryptocurrency

This is pretty horrific:

…a group of men behind a violent crime spree designed to compel victims to hand over access to their cryptocurrency savings. That announcement and the criminal complaint laying out charges against St. Felix focused largely on a single theft of cryptocurrency from an elderly North Carolina couple, whose home St. Felix and one of his accomplices broke into before physically assaulting the two victims—­both in their seventies—­and forcing them to transfer more than $150,000 in Bitcoin and Ether to the thieves’ crypto wallets.

I think cryptocurrencies are more susceptible to this kind of real-world attack because they are largely outside the conventional banking system. Yet another reason to stay away from them.

Posted on July 18, 2024 at 11:33 AM • View Comments

The Insecurity of Video Doorbells

Consumer Reports has analyzed a bunch of popular Internet-connected video doorbells. Their security is terrible.

First, these doorbells expose your home IP address and WiFi network name to the internet without encryption, potentially opening your home network to online criminals.

[…]

Anyone who can physically access one of the doorbells can take over the device—no tools or fancy hacking skills needed.

Posted on March 5, 2024 at 7:05 AM • View Comments

Museum Security

Interesting interview:

Banks don’t take millions of dollars and put them in plastic bags and hang them on the wall so everybody can walk right up to them. But we do basically the same thing in museums and hang the assets right out on the wall. So it’s our job, then, to either use technology or develop technology that protects the art, to hire honest guards that are trainable and able to meet the challenge and alert and so forth. And we have to keep them alert because it’s the world’s most boring job. It might be great for you to go to a museum and see it for a day, but they stand in that same gallery year after year, and so they get mental fatigue. And so we have to rotate them around and give them responsibilities that keep them stimulated and keep them fresh.

It’s a challenge. But we try to predict the items that might be most vulnerable. Which are not necessarily most valuable; some things have symbolic significance to them. And then we try to predict what the next targets might be and advise our clients that they maybe need to put special security on those items.

Posted on October 19, 2022 at 6:16 AM • View Comments

Determining Key Shape from Sound

It’s not yet very accurate or practical, but under ideal conditions it is possible to figure out the shape of a house key by listening to it being used.

Listen to Your Key: Towards Acoustics-based Physical Key Inference

Abstract: Physical locks are one of the most prevalent mechanisms for securing objects such as doors. While many of these locks are vulnerable to lock-picking, they are still widely used as lock-picking requires specific training with tailored instruments, and easily raises suspicion. In this paper, we propose SpiKey, a novel attack that significantly lowers the bar for an attacker as opposed to the lock-picking attack, by requiring only the use of a smartphone microphone to infer the shape of victim’s key, namely bittings(or cut depths) which form the secret of a key. When a victim inserts his/her key into the lock, the emitted sound is captured by the attacker’s microphone.SpiKey leverages the time difference between audible clicks to ultimately infer the bitting information, i.e., shape of the physical key. As a proof-of-concept, we provide a simulation, based on real-world recordings, and demonstrate a significant reduction in search spacefrom a pool of more than 330 thousand keys to three candidate keys for the most frequent case.

Scientific American podcast:

The strategy is a long way from being viable in the real world. For one thing, the method relies on the key being inserted at a constant speed. And the audio element also poses challenges like background noise.

Boing Boing post.

EDITED TO ADD (4/14): I seem to have blogged this previously.

Posted on March 24, 2021 at 6:10 AM • View Comments

Attack against Florida Water Treatment Facility

A water treatment plant in Oldsmar, Florida, was attacked last Friday. The attacker took control of one of the systems, and increased the amount of sodium hydroxide—that’s lye—by a factor of 100. This could have been fatal to people living downstream, if an alert operator hadn’t noticed the change and reversed it.

We don’t know who is behind this attack. Despite its similarities to a Russian attack of a Ukrainian power plant in 2015, my bet is that it’s a disgruntled insider: either a current or former employee. It just doesn’t make sense for Russia to be behind this.

ArsTechnica is reporting on the poor cybersecurity at the plant:

The Florida water treatment facility whose computer system experienced a potentially hazardous computer breach last week used an unsupported version of Windows with no firewall and shared the same TeamViewer password among its employees, government officials have reported.

Brian Krebs points out that the fact that we know about this attack is what’s rare:

Spend a few minutes searching Twitter, Reddit or any number of other social media sites and you’ll find countless examples of researchers posting proof of being able to access so-called “human-machine interfaces”—basically web pages designed to interact remotely with various complex systems, such as those that monitor and/or control things like power, water, sewage and manufacturing plants.

And yet, there have been precious few known incidents of malicious hackers abusing this access to disrupt these complex systems. That is, until this past Monday, when Florida county sheriff Bob Gualtieri held a remarkably clear-headed and fact-filled news conference about an attempt to poison the water supply of Oldsmar, a town of around 15,000 not far from Tampa.

Posted on February 12, 2021 at 6:08 AM • View Comments

Including Hackers in NATO Wargames

This essay makes the point that actual computer hackers would be a useful addition to NATO wargames:

The international information security community is filled with smart people who are not in a military structure, many of whom would be excited to pose as independent actors in any upcoming wargames. Including them would increase the reality of the game and the skills of the soldiers building and training on these networks. Hackers and cyberwar experts would demonstrate how industrial control systems such as power supply for refrigeration and temperature monitoring in vaccine production facilities are critical infrastructure; they’re easy targets and should be among NATO’s priorities at the moment.

Diversity of thought leads to better solutions. We in the information security community strongly support the involvement of acknowledged nonmilitary experts in the development and testing of future cyberwar scenarios. We are confident that independent experts, many of whom see sharing their skills as public service, would view participation in these cybergames as a challenge and an honor.

Posted on January 29, 2021 at 12:03 PM • View Comments

On US Capitol Security—By Someone Who Manages Arena-Rock-Concert Security

Smart commentary:

…I was floored on Wednesday when, glued to my television, I saw police in some areas of the U.S. Capitol using little more than those same mobile gates I had ­ the ones that look like bike racks that can hook together ­ to try to keep the crowds away from sensitive areas and, later, push back people intent on accessing the grounds. (A new fence that appears to be made of sturdier material was being erected on Thursday.) That’s the same equipment and approximately the same amount of force I was able to use when a group of fans got a little feisty and tried to get backstage at a Vanilla Ice show.

[…]

There’s not ever going to be enough police or security at any event to stop people if they all act in unison; if enough people want to get to Vanilla Ice at the same time, they’re going to get to Vanilla Ice. Social constructs and basic decency, not lightweight security gates, are what hold everyone except the outliers back in a typical crowd.

[…]

When there are enough outliers in a crowd, it throws the normal dynamics of crowd control off; everyone in my business knows this. Citizens tend to hold each other to certain standards ­ which is why my 40,000-person town does not have 40,000 police officers, and why the 8.3 million people of New York City aren’t policed by 8.3 million police officers.

Social norms are the fabric that make an event run smoothly—and, really, hold society together. There aren’t enough police in your town to handle it if everyone starts acting up at the same time.

I like that she uses the term “outliers,” and I make much the same points in Liars and Outliers.

Posted on January 13, 2021 at 6:06 AM • View Comments

1 2 3 … 25

Sidebar photo of Bruce Schneier by Joe MacInnis.