Security Risks of New .zip and .mov Domains
Researchers are worried about Google’s .zip and .mov domains, because they are confusing. Mistaking a URL for a filename could be a security vulnerability.
Page 1 of 22
Researchers are worried about Google’s .zip and .mov domains, because they are confusing. Mistaking a URL for a filename could be a security vulnerability.
Another nation-state malware, Russian in origin:
In the early stages of the war in Ukraine in 2022, PIPEDREAM, a known malware was quietly on the brink of wiping out a handful of critical U.S. electric and liquid natural gas sites. PIPEDREAM is an attack toolkit with unmatched and unprecedented capabilities developed for use against industrial control systems (ICSs).
The malware was built to manipulate the network communication protocols used by programmable logic controllers (PLCs) leveraged by two critical producers of PLCs for ICSs within the critical infrastructure sector, Schneider Electric and OMRON.
Stanford and Georgetown have a new report on the security risks of AI—particularly adversarial machine learning—based on a workshop they held on the topic.
Jim Dempsey, one of the workshop organizers, wrote a blog post on the report:
As a first step, our report recommends the inclusion of AI security concerns within the cybersecurity programs of developers and users. The understanding of how to secure AI systems, we concluded, lags far behind their widespread adoption. Many AI products are deployed without institutions fully understanding the security risks they pose. Organizations building or deploying AI models should incorporate AI concerns into their cybersecurity functions using a risk management framework that addresses security throughout the AI system life cycle. It will be necessary to grapple with the ways in which AI vulnerabilities are different from traditional cybersecurity bugs, but the starting point is to assume that AI security is a subset of cybersecurity and to begin applying vulnerability management practices to AI-based features. (Andy Grotto and I have vigorously argued against siloing AI security in its own governance and policy vertical.)
Our report also recommends more collaboration between cybersecurity practitioners, machine learning engineers, and adversarial machine learning researchers. Assessing AI vulnerabilities requires technical expertise that is distinct from the skill set of cybersecurity practitioners, and organizations should be cautioned against repurposing existing security teams without additional training and resources. We also note that AI security researchers and practitioners should consult with those addressing AI bias. AI fairness researchers have extensively studied how poor data, design choices, and risk decisions can produce biased outcomes. Since AI vulnerabilities may be more analogous to algorithmic bias than they are to traditional software vulnerabilities, it is important to cultivate greater engagement between the two communities.
Another major recommendation calls for establishing some form of information sharing among AI developers and users. Right now, even if vulnerabilities are identified or malicious attacks are observed, this information is rarely transmitted to others, whether peer organizations, other companies in the supply chain, end users, or government or civil society observers. Bureaucratic, policy, and cultural barriers currently inhibit such sharing. This means that a compromise will likely remain mostly unnoticed until long after attackers have successfully exploited vulnerabilities. To avoid this outcome, we recommend that organizations developing AI models monitor for potential attacks on AI systems, create—formally or informally—a trusted forum for incident information sharing on a protected basis, and improve transparency.
Researchers at Russian cybersecurity firm Kaspersky today revealed that they identified a small number of cryptocurrency-focused firms as at least some of the victims of the 3CX software supply-chain attack that’s unfolded over the past week. Kaspersky declined to name any of those victim companies, but it notes that they’re based in “western Asia.”
Security firms CrowdStrike and SentinelOne last week pinned the operation on North Korean hackers, who compromised 3CX installer software that’s used by 600,000 organizations worldwide, according to the vendor. Despite the potentially massive breadth of that attack, which SentinelOne dubbed “Smooth Operator,” Kaspersky has now found that the hackers combed through the victims infected with its corrupted software to ultimately target fewer than 10 machines—at least as far as Kaspersky could observe so far—and that they seemed to be focusing on cryptocurrency firms with “surgical precision.”
Both Google’s Pixel’s Markup Tool and the Windows Snipping Tool have vulnerabilities that allow people to partially recover content that was edited out of images.
EDITED TO ADD (4/14): Steven Murdoch has a good explanation as to why this happened—and to two very different snipping tools.
OpenAI has disabled ChatGPT’s privacy history, almost certainly because it had a security flaw where users were seeing each others’ histories.
Last week, the Biden administration released a new National Cybersecurity Strategy (summary here). There is lots of good commentary out there. It’s basically a smart strategy, but the hard parts are always the implementation details. It’s one thing to say that we need to secure our cloud infrastructure, and another to detail what the means technically, who pays for it, and who verifies that it’s been done.
One of the provisions getting the most attention is a move to shift liability to software vendors, something I’ve been advocating for since at least 2003.
What will it take for policy makers to take cybersecurity seriously? Not minimal-change seriously. Not here-and-there seriously. But really seriously. What will it take for policy makers to take cybersecurity seriously enough to enact substantive legislative changes that would address the problems? It’s not enough for the average person to be afraid of cyberattacks. They need to know that there are engineering fixes—and that’s something we can provide.
For decades, I have been waiting for the “big enough” incident that would finally do it. In 2015, Chinese military hackers hacked the Office of Personal Management and made off with the highly personal information of about 22 million Americans who had security clearances. In 2016, the Mirai botnet leveraged millions of Internet-of-Things devices with default admin passwords to launch a denial-of-service attack that disabled major Internet platforms and services in both North America and Europe. In 2017, hackers—years later we learned that it was the Chinese military—hacked the credit bureau Equifax and stole the personal information of 147 million Americans. In recent years, ransomware attacks have knocked hospitals offline, and many articles have been written about Russia inside the U.S. power grid. And last year, the Russian SVR hacked thousands of sensitive networks inside civilian critical infrastructure worldwide in what we’re now calling Sunburst (and used to call SolarWinds).
Those are all major incidents to security people, but think about them from the perspective of the average person. Even the most spectacular failures don’t affect 99.9% of the country. Why should anyone care if the Chinese have his or her credit records? Or if the Russians are stealing data from some government network? Few of us have been directly affected by ransomware, and a temporary Internet outage is just temporary.
Cybersecurity has never been a campaign issue. It isn’t a topic that shows up in political debates. (There was one question in a 2016 Clinton–Trump debate, but the response was predictably unsubstantive.) This just isn’t an issue that most people prioritize, or even have an opinion on.
So, what will it take? Many of my colleagues believe that it will have to be something with extreme emotional intensity—sensational, vivid, salient—that results in large-scale loss of life or property damage. A successful attack that actually poisons a water supply, as someone tried to do in January by raising the levels of lye at a Florida water-treatment plant. (That one was caught early.) Or an attack that disables Internet-connected cars at speed, something that was demonstrated by researchers in 2014. Or an attack on the power grid, similar to what Russia did to the Ukraine in 2015 and 2016. Will it take gas tanks exploding and planes falling out of the sky for the average person to read about the casualties and think “that could have been me”?
Here’s the real problem. For the average nonexpert—and in this category I include every lawmaker—to push for change, they not only need to believe that the present situation is intolerable, they also need to believe that an alternative is possible. Real legislative change requires a belief that the never-ending stream of hacks and attacks is not inevitable, that we can do better. And that will require creating working examples of secure, dependable, resilient systems.
Providing alternatives is how engineers help facilitate social change. We could never have eliminated sales of tungsten-filament household light bulbs if fluorescent and LED replacements hadn’t become available. Reducing the use of fossil fuel for electricity generation requires working wind turbines and cost-effective solar cells.
We need to demonstrate that it’s possible to build systems that can defend themselves against hackers, criminals, and national intelligence agencies; secure Internet-of-Things systems; and systems that can reestablish security after a breach. We need to prove that hacks aren’t inevitable, and that our vulnerability is a choice. Only then can someone decide to choose differently. When people die in a cyberattack and everyone asks “What can be done?” we need to have something to tell them.
We don’t yet have the technology to build a truly safe, secure, and resilient Internet and the computers that connect to it. Yes, we have lots of security technologies. We have older secure systems—anyone still remember Apollo’s DomainOS and MULTICS?—that lost out in a market that didn’t reward security. We have newer research ideas and products that aren’t successful because the market still doesn’t reward security. We have even newer research ideas that won’t be deployed, again, because the market still prefers convenience over security.
What I am proposing is something more holistic, an engineering research task on a par with the Internet itself. The Internet was designed and built to answer this question: Can we build a reliable network out of unreliable parts in an unreliable world? It turned out the answer was yes, and the Internet was the result. I am asking a similar research question: Can we build a secure network out of insecure parts in an insecure world? The answer isn’t obviously yes, but it isn’t obviously no, either.
While any successful demonstration will include many of the security technologies we know and wish would see wider use, it’s much more than that. Creating a secure Internet ecosystem goes beyond old-school engineering to encompass the social sciences. It will include significant economic, institutional, and psychological considerations that just weren’t present in the first few decades of Internet research.
Cybersecurity isn’t going to get better until the economic incentives change, and that’s not going to change until the political incentives change. The political incentives won’t change until there is political liability that comes from voter demands. Those demands aren’t going to be solely the results of insecurity. They will also be the result of believing that there’s a better alternative. It is our task to research, design, build, test, and field that better alternative—even though the market couldn’t care less right now.
This essay originally appeared in the May/June 2021 issue of IEEE Security & Privacy. I forgot to publish it here.
The field of machine learning (ML) security—and corresponding adversarial ML—is rapidly advancing as researchers develop sophisticated techniques to perturb, disrupt, or steal the ML model or data. It’s a heady time; because we know so little about the security of these systems, there are many opportunities for new researchers to publish in this field. In many ways, this circumstance reminds me of the cryptanalysis field in the 1990. And there is a lesson in that similarity: the complex mathematical attacks make for good academic papers, but we mustn’t lose sight of the fact that insecure software will be the likely attack vector for most ML systems.
We are amazed by real-world demonstrations of adversarial attacks on ML systems, such as a 3D-printed object that looks like a turtle but is recognized (from any orientation) by the ML system as a gun. Or adding a few stickers that look like smudges to a stop sign so that it is recognized by a state-of-the-art system as a 45 mi/h speed limit sign. But what if, instead, somebody hacked into the system and just switched the labels for “gun” and “turtle” or swapped “stop” and “45 mi/h”? Systems can only match images with human-provided labels, so the software would never notice the switch. That is far easier and will remain a problem even if systems are developed that are robust to those adversarial attacks.
At their core, modern ML systems have complex mathematical models that use training data to become competent at a task. And while there are new risks inherent in the ML model, all of that complexity still runs in software. Training data are still stored in memory somewhere. And all of that is on a computer, on a network, and attached to the Internet. Like everything else, these systems will be hacked through vulnerabilities in those more conventional parts of the system.
This shouldn’t come as a surprise to anyone who has been working with Internet security. Cryptography has similar vulnerabilities. There is a robust field of cryptanalysis: the mathematics of code breaking. Over the last few decades, we in the academic world have developed a variety of cryptanalytic techniques. We have broken ciphers we previously thought secure. This research has, in turn, informed the design of cryptographic algorithms. The classified world of the NSA and its foreign counterparts have been doing the same thing for far longer. But aside from some special cases and unique circumstances, that’s not how encryption systems are exploited in practice. Outside of academic papers, cryptosystems are largely bypassed because everything around the cryptography is much less secure.
I wrote this in my book, Data and Goliath:
The problem is that encryption is just a bunch of math, and math has no agency. To turn that encryption math into something that can actually provide some security for you, it has to be written in computer code. And that code needs to run on a computer: one with hardware, an operating system, and other software. And that computer needs to be operated by a person and be on a network. All of those things will invariably introduce vulnerabilities that undermine the perfection of the mathematics…
This remains true even for pretty weak cryptography. It is much easier to find an exploitable software vulnerability than it is to find a cryptographic weakness. Even cryptographic algorithms that we in the academic community regard as “broken”—meaning there are attacks that are more efficient than brute force—are usable in the real world because the difficulty of breaking the mathematics repeatedly and at scale is much greater than the difficulty of breaking the computer system that the math is running on.
ML systems are similar. Systems that are vulnerable to model stealing through the careful construction of queries are more vulnerable to model stealing by hacking into the computers they’re stored in. Systems that are vulnerable to model inversion—this is where attackers recover the training data through carefully constructed queries—are much more vulnerable to attacks that take advantage of unpatched vulnerabilities.
But while security is only as strong as the weakest link, this doesn’t mean we can ignore either cryptography or ML security. Here, our experience with cryptography can serve as a guide. Cryptographic attacks have different characteristics than software and network attacks, something largely shared with ML attacks. Cryptographic attacks can be passive. That is, attackers who can recover the plaintext from nothing other than the ciphertext can eavesdrop on the communications channel, collect all of the encrypted traffic, and decrypt it on their own systems at their own pace, perhaps in a giant server farm in Utah. This is bulk surveillance and can easily operate on this massive scale.
On the other hand, computer hacking has to be conducted one target computer at a time. Sure, you can develop tools that can be used again and again. But you still need the time and expertise to deploy those tools against your targets, and you have to do so individually. This means that any attacker has to prioritize. So while the NSA has the expertise necessary to hack into everyone’s computer, it doesn’t have the budget to do so. Most of us are simply too low on its priorities list to ever get hacked. And that’s the real point of strong cryptography: it forces attackers like the NSA to prioritize.
This analogy only goes so far. ML is not anywhere near as mathematically sound as cryptography. Right now, it is a sloppy misunderstood mess: hack after hack, kludge after kludge, built on top of each other with some data dependency thrown in. Directly attacking an ML system with a model inversion attack or a perturbation attack isn’t as passive as eavesdropping on an encrypted communications channel, but it’s using the ML system as intended, albeit for unintended purposes. It’s much safer than actively hacking the network and the computer that the ML system is running on. And while it doesn’t scale as well as cryptanalytic attacks can—and there likely will be a far greater variety of ML systems than encryption algorithms—it has the potential to scale better than one-at-a-time computer hacking does. So here again, good ML security denies attackers all of those attack vectors.
We’re still in the early days of studying ML security, and we don’t yet know the contours of ML security techniques. There are really smart people working on this and making impressive progress, and it’ll be years before we fully understand it. Attacks come easy, and defensive techniques are regularly broken soon after they’re made public. It was the same with cryptography in the 1990s, but eventually the science settled down as people better understood the interplay between attack and defense. So while Google, Amazon, Microsoft, and Tesla have all faced adversarial ML attacks on their production systems in the last three years, that’s not going to be the norm going forward.
All of this also means that our security for ML systems depends largely on the same conventional computer security techniques we’ve been using for decades. This includes writing vulnerability-free software, designing user interfaces that help resist social engineering, and building computer networks that aren’t full of holes. It’s the same risk-mitigation techniques that we’ve been living with for decades. That we’re still mediocre at it is cause for concern, with regard to both ML systems and computing in general.
I love cryptography and cryptanalysis. I love the elegance of the mathematics and the thrill of discovering a flaw—or even of reading and understanding a flaw that someone else discovered—in the mathematics. It feels like security in its purest form. Similarly, I am starting to love adversarial ML and ML security, and its tricks and techniques, for the same reasons.
I am not advocating that we stop developing new adversarial ML attacks. It teaches us about the systems being attacked and how they actually work. They are, in a sense, mechanisms for algorithmic understandability. Building secure ML systems is important research and something we in the security community should continue to do.
There is no such thing as a pure ML system. Every ML system is a hybrid of ML software and traditional software. And while ML systems bring new risks that we haven’t previously encountered, we need to recognize that the majority of attacks against these systems aren’t going to target the ML part. Security is only as strong as the weakest link. As bad as ML security is right now, it will improve as the science improves. And from then on, as in cryptography, the weakest link will be in the software surrounding the ML system.
This essay originally appeared in the May 2020 issue of IEEE Computer. I forgot to reprint it here.
NIST is planning a significant update of its Cybersecurity Framework. At this point, it’s asking for feedback and comments to its concept paper.
- Do the proposed changes reflect the current cybersecurity landscape (standards, risks, and technologies)?
- Are the proposed changes sufficient and appropriate? Are there other elements that should be considered under each area?
- Do the proposed changes support different use cases in various sectors, types, and sizes of organizations (and with varied capabilities, resources, and technologies)?
- Are there additional changes not covered here that should be considered?
- For those using CSF 1.1, would the proposed changes affect continued adoption of the Framework, and how so?
- For those not using the Framework, would the proposed changes affect the potential use of the Framework?
The NIST Cybersecurity Framework has turned out to be an excellent resource. If you use it at all, please help with version 2.0.
EDITED TO ADD (2/14): Details on progress and how to engage.
Sidebar photo of Bruce Schneier by Joe MacInnis.