Unsupervised Learning

Daniel Miessler

Unsupervised Learning is about ideas and trends in Cybersecurity, National Security, AI, Technology, and Culture—and how best to upgrade ourselves to be ready for what's coming.

Subscribe Listen in podnod RSS feed Show website

545 episodes
Updated July 9
Explicit

Episodes545

Newest first

Sep 2, 2019 · 35 min
Unsupervised Learning: No. 192
Ring has already partnered with over 400 police departments. As you know, I'm torn on this kind of tech. Neighborhood watch can be a good thing, and it can also be a bad thing. Technology tends to magnify both weaknesses and strengths, so it can make neighborhood watch really great, or it can turn it into a nightmare. The problem is that you can easily start on the positive side, build it all the way up, and then in a few legal, policy, and tech changes have it turn into the oppressive form. Some say this is a reason not to do any of this stuff, but I disagree. We know someone is going to do it, so I think the best thing that can be done is to build a benign version and hope it wins in the market. More People are drawing comparisons between China's social credit system (which is actually multiple systems) and the Silicon Valley's various apps that have internal rating systems. They're saying that these ratings will eventually be used to make decisions about things that matter. Sure, but this has existed throughout human history. Word of mouth, blacklists, etc.: these are all ways of extending the reach of good or bad reputation. I think whenever someone points out the downside of a technology, we should ask ourselves whether that dynamic exists already in the real world, and adjust our opinions accordingly. More The Pentagon is worried that China will beat the US in AI if we don't create a stronger link between the government and both academia and industry, which China is good at. We basically need to move faster from edge concepts to practical implementations, but it's damn hard to do this when we have all sorts of legal and ethical constraints that China doesn't have. Our caution and morality are a definite weakness in this case. More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Aug 26, 2019 · 25 min
Unsupervised Learning: No. 191
Protestors in Hong Kong are physically attacking and destroying facial recognition cameras. More Palo Alto says 7 out of 10 new domain registrations (NDRs) are either malicious or not safe for work, and they encourage companies to block them. More Lt. Gen. Fogarty is fighting to change the name of Army Cyber Command to Army Information Warfare Command, and to give the group a much larger scope in its mission. More We continue to see attacks against open source supply chains, in packages like NPM, RubyGems, Webmin, and many others. It's about to become imperative for people to understand—and to be able to validate—the entire chain of trust that a given application sits upon before they use it. There have been many companies in this space in the past, but I expect to see them (and new players) get a lot more attention soon. More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Aug 19, 2019 · 5 min
The Difference Between Data, Information, and Intelligence
The terms intelligence, information, and data are thrown around pretty loosely in most tech circles, and this inevitably leads to people confusing and/or conflating them. What follows is a simple explanation of how the related terms are different from each other, and how they work together. Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Aug 19, 2019 · 22 min
Unsupervised Learning: No. 190
There are some seriously nasty Windows RDP bugs out there. If you have RDP facing the internet, make sure you're patched. And try to get to VPN as soon as possible. More A huge survey of firmware security has found virtually no improvement over the last 15 years. People seem surprised by this, but it is exactly what I would have predicted based on my analysis here. Basically, for most people not in the industry, our current state is actually fine. More NYPD has over 82K peoples' DNA in a database, and the program has little visibility and oversight. More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Aug 13, 2019 · 8 min
Unsupervised Learning: No. 189
Ring is developing two-way relationships with hundreds of police departments in the US. This allows Ring users to be alerted to crime in their area via 911 data, and police departments to pull video from participating Ring devices. This is the type of functionality that most people will see and think, “Wow, I'd love to have that!”, which is why it's going to be very successful. But it's also one tiny step away from something terrifying. More A number of critical bugs in VxWorks are going to cause issues with infrastructure for years to come. More DARPA is building a $10 million dollar, open source voting system with a focus on security. More It looks like China's social credit system might not be a giant monolithic system, but rather a series of siloed experiments. More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jul 29, 2019 · 19 min
Unsupervised Learning: No. 188
Marcus Hutchins got off with time-served, and people have feelings. The range basically goes from 'he did nothing wrong', to, 'he should rot in prison'. In my mind this outcome was close to perfect. Remember, he went through two years of hell since being brought up charges, he's still a convicted felon, and he also is largely banned from the US. I think it's good that he admitted guilt, faced consequences, and is being offered a chance to continue giving back to the community. More Attorney General Barr said recently that companies should put backdoors in their products that bypass encryption, or else the government will pass laws that require it. This is unspeakably stupid. Without even getting into the philosophy of whether the internet can host a private conversation (which requires a warrant to tap), we can just start with the fact that backdoors present a clear and present danger to security, right now, due to the weaknesses of those who create them. If the NSA can be hacked or somehow lose its sensitive tools and materials, there's no company this cannot happen to. Purposefully installing backdoors therefore equates (effectively) to giving such access to attackers. Unacceptable. More Equifax is offering people $125 dollars in reparations for them losing all your data. But to get it, you have to log in and give a bunch of data about yourself. It's hilarious. They made money offering credit protection after the breach, and now they're going to collect updated information on anyone who wants to collect $125. On Twitter I called this a sadder and more permanent form of giving plasma. More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jul 26, 2019 · 7 min
Humans Are Genebots
Unpacking the evolution-granted bliss of prep schools and elite institutions, and why they resonate so much with us. Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jul 25, 2019 · 8 min
Machine Learning Doesn’t Introduce Unfairness—It Reveals It
The difference between unfairness and bias in machine learning. Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jul 22, 2019 · 35 min
Unsupervised Learning: No. 187
Lots of people in the security community went silly over the FaceApp application last week, basically saying that you shouldn't be using the application because they'll steal your face and then be able to impersonate you. Oh, and then it turned out to be a Russian company who put out the application, and that made it 100x worse. The problem here is the lack of Threat Model Thinking. When it comes to election security, propaganda discussions, etc., I am quite concerned about Putin's willingness and ability to harm our country's cohesion through memes and social media. But that does not extend to some random company stealing faces. Why? Because before you can get legitimately concerned about something, you have to be able to describe a threat scenario in which that thing becomes dangerous. As I talked about in this piece, pictures of your face are not the same as your face when it comes to biometric authentication. There's a reason companies need a specific device, combined with their custom algorithm, in order to enroll you in a facial identification system. They scan you in a very specific way and then store your data (which is just a representation, not your actual face) in a very specific way. Then they need to use that same exact system to scan you again, so they can compare the two representations to each other. That isn't happening with random apps that have pictures of you. And even if that were the case, they could just get your face off your social media, where those same people who are worried are more than happy to take selfies, put their pictures on profile pictures, and make sure as many people see them as possible. There are actual negative things that can be done with images (like making Deepfakes of you), and that will get easier over time, but the defense for that is to have zero pictures of you…anywhere. And once again you have to ask who would be doing that to you, and why. Bottom line: authentication systems take special effort to try to ensure that the input given is the same as the enrollment item, e.g., (face, fingerprint, etc.), so it will not be easy any time soon to go from a random picture to something that can full a face scanner or fingerprint reader at the airport. People reading this probably already know this, but spread the word: threat modeling is one of our best tools for removing emotion from risk management. A contractor named SyTech that does work with Russian FSB has been breached, resulting in the release of 7.5TB of data on the FSB's various projects. This is obviously embarrassing for SyTech and the FSB, but the leaked projects focused on de-anonymization, spying on Russian businesses, and the project to break Russia away from the Internet, which are all known and expected efforts. So there don't seem to be any big reveals as a result of the leak. More Someone discovered that a bunch of browser extensions were reading things they shouldn't be, and sending them out to places they shouldn't be. This is not surprising to me. Chrome extensions are like Android apps, which should tell you all you need to know about installing random ones that seem interesting. My policy on browser extensions is extremely strict for this reason. People need to understand how insane the entire idea of the modern web is. We're visiting URLs that are executing code on our machines. And not just code from that website, but code from thousands of other websites in an average browsing session. It's a garbage fire. And the only defense really is to question how much you trust your browser, your operating system, and the original site you're visiting. But even then you're still exposing yourself to significant and continuously-evolving risk when you run around clicking things online. And the worst possible thing you can do in this situation is install more functionality, which gives more parties, more access, to that giant stack of assumptions you're making just by using a web browser. The best possible stance is to have as few people possible with access to your particular dumpster. And that means installing as few highly-vetted add-ons as possible. More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jul 20, 2019 · 5 min
Time Speeds Up When You’re Wasting It
An essay on why time can feel like it's speeding up when you get older, and how to slow it back down. Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jul 15, 2019 · 20 min
Unsupervised Learning: No. 186
Parts of Manhattan had a power outage Saturday night, which happened to be the anniversary of another power outage in 1977. The power company apologized but didn't explain what happened. The hacker in me thinks this could easily be a probing shot by a sophisticated attacker, or a fun prank by amateurs. But the overwhelming odds are on simple failure. Either way, this country needs to get a whole lot more resilient to small attacks, because enough small ones can quickly become a big one. More Zoom has had a bad week or two. Not only did it have a major vuln, but it turned out to be part of the design, and they moved relatively slowly in addressing it, and then companies started auto-uninstalling it from their OS. They had a lot of momentum going in the space, too. This will sting for sure. More Facebook will be fined $5 billion over its various privacy catastrophes. More Marriott is being fined $124 million over the Starwood breach. Real question: how does that compare to their coffee budget? More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jul 8, 2019 · 21 min
Unsupervised Learning: No. 185
The Telegraph has found strong links between Huawei employees and Chinese intelligence agencies. The Huawei counter was that this was extremely common among telecom companies, and that it wasn't a big deal. The counter to that counter was, basically, "Well, then why did you try to hide it?" /gg More The NPM security team caught a malicious package designed to steal cryptocurrency. A lot of these packages work by uploading something useful, waiting until it's used by lots of people, and then updating it to have the malicious payload. My buddy Andre Eleuterio did the IR on the situation there at NPM, and said they're constantly improving their ability to detect these kinds of attacks. Luckily NPM's security team had the talent and tooling to detect such a thing, but think of how many similar companies aren't so equipped. I think any team that's part of a supply chain should be thinking about this type of attack very seriously. More Federal agents are mining state DMV photos to feed their facial recognition systems, and they're doing it without proper authorizations or consent. To me this has always been inevitable because—as Benedict Evans pointed out—it's a natural extension of what humans already do. You already have wanted posters. You already have known suspects lists. And it's already ok for any citizen or any cop to see any person on that list and report them. In fact it's not just possible, it's encouraged. So the only thing happening here is that process is becoming a whole lot more aware (through more sensors), and therefore more effective. Of course, any broken algorithms that identify the wrong people, or automatically single out groups of people without actual matches, those issues need to be snuffed out for sure. But we can't expect society to not use superior machine alternatives to existing human processes, such as identifying suspects in public. That just isn't realistic. Our role as security people should be making sure these systems are as accurate as possible, with as little bias as possible, by the best possible people. In other words, we should spend our cycles improving reality, not trying to stop it from happening. More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jul 3, 2019 · 8 min
The World is Collapsing Into Two Countries—Green and Red
The world being sorted into two different countries—a Green country of the top 10% of income/wealk, and a Red country that's everyone else. These countries are separated not by geography, but by class. Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jul 1, 2019 · 18 min
Unsupervised Learning: No. 184
I created a new tutorial on OWASP Amass, and just joined the team as a contributor as well. Tutorial Chinese hacking groups have been embedded deep inside multiple major US tech firms for many years, including Fujitsu, Tata, NTT, Dimension Data, and HPE. The first thing you should be thinking is where else they are today. More Amazon is getting heavier into the SIEM space (and perhaps others) with their new Amazon Security Hub offering. It takes in lots of event types from various AWS services, and surfaces what it thinks is most important. Of course, it doesn't do this for other product types, i.e., non-AWS stuff, but that could come eventually. More Amazon also launched a new service that lets you monitor your AWS VPC traffic. And lots of vendors are announcing their support for it. More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jun 24, 2019 · 13 min
Unsupervised Learning: No. 183
There's a Linux vulnerability called SACK Panic (among other names) that takes advantage of a kernel feature called Selective ACK. The feature lets systems tell the other side of the conversation how much data it's received, and it turns out it can be overflowed or fuzzed. The former creates a crash, and the latter creates a slowdown. You should patch. And if you have any services facing the internet running Linux, you should definitely patch. More A Florida city paid $600,000 in bitcoin to get access to their data back from a ransomware gang. More Magic Leap is suing former engineer Chi Xu for allegedly using his knowledge of the headset to make a version for China. More The average security group is running over 50 security tools. As my friend Jeremiah once said when looking at a Momentum Partners slide, "Are we secure yet?" More Amazon just got a patent for using delivery drones for surveillance. I don't necessarily think that means they'll use delivery drones for surveillance though. That's what a lot of the conspiracy theorists will say, though—just based on them getting a patent for using delivery drones for surveillance. Actually, the patent is a bit more benign than my joke implies. It's designed to monitor opted-in people's property, a lot like a house camera or a Ring device. Makes sense. But still. More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jun 18, 2019 · 10 min
Unsupervised Learning: No. 182
The US is supposedly ramping up attacks against Russian power grid through the use of new cyberattack powers granted by Trump. I am happy to hear of this, but it's an example of where we as outsiders can only know a tiny fragment of the story. But any signs that this administration sees Russia as a foe, and are treating it as such, are positive in my view. More Adobe is entering the deepfakes arena by showing off research tools designed to detect manipulated photos. More Target stores have been hit by major outages. More Many places are using very granular bluetooth beacon tracking to watch you move throughout their businesses, including airports, malls, subways, buses, gyms, hotels, festivals, museums, etc. More The US is going after ethnic Chinese researchers in the medical field, and specifically at cancer centers. I'm all for becoming more aggressive towards the Chinese government pilfering the world's intellectual property, but, um, cancer research is one thing that I think it's ok to spread widely. It's not like they're stealing the only copy of the research; they're just sharing it. Maybe I'm missing something, but if that something is just about who makes the profit, then I'm calling Meh. More Firewalling outbound DNS could save companies billions. Yes! I've been on about this for years. More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jun 11, 2019 · 24 min
Unsupervised Learning: No. 181
Some absolutely fascinating research has just come out on what percentages and types of vulnerabilities are actually exploited in the wild. It found that only 5.5% of vulnerabilities discovered between 2009 and 2018 were actually exploited, with most of those being issues with a CVSS score of 9 or 10. The best part of the paper, however, was a discussion of optimal patching strategies, where they looked at different methodologies for what to patch and measured them against each other based on coverage (no misses) and efficiency (not patching what you don't have to). Options included patching by CVSS, whether or not there are public exploits, by vulnerability tags, etc. The ML model performed best, but it seemed that patching the CVSS 7 and above was decent as well, and for more efficiency but less coverage—CVSS 9 and above. Super interesting paper. More The US is going to start requiring 5 years of social media account history from Visa applicants, as part of the filtering process. I'm genuinely curious as to how effective this is going to be. On the one hand, there will now be a market for creating and maintaining fake social media accounts that people can use for this purpose. But on the other hand, there will be many who don't want to go to that effort and either won't try to come, or will get caught in the filter. As with most things, the efficacy will come down to execution. More A team at Stanford has made it possible to edit video using a text editor. So, editing the things that were said by the actual subject, to say something else entirely, but having it seamlessly injected into the video so it looks completely natural. More Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jun 8, 2019 · 6 min
Grit is the Ultimate Privilege
An argument that we should acknowledge grit as one of the most powerful causal factors in success, and figure out ways to bring its benefits to everyone. Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
Jun 6, 2019 · 4 min
Why Software Remains Insecure
A concise explanation of why software continues to have security and quality problems after decades of supposedly trying to address the problem. Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.
May 28, 2019 · 17 min
Unsupervised Learning: No. 179
The Deepfakes thing is already starting to have an impact, and it didn't even involve actual Deepfake (GAN ML) technology. A video was spread of Nancy Pelosi speaking very slowly and seeming to stumble over her words, which made her look quite bad. The video was virally shared throughout social media on the right. Problem is, it was intentionally slowed down to make her look old/stupid/crazy. What this shows us is that it's not the machine learning that makes Deepfakes dangerous; it's the willingness of a massive percentage of the US population to believe total garbage without an ounce of scrutiny. It doesn't matter if Deepfakes can be shown to be fake because people are matching evidence to their emotions, not the other way around. The vulnerability is our ignorance and cynicism, not a spoofing technology. And as I wrote about a couple of years ago, this will be used as a weapon against us. More Essay A real estate insurance website for First American Financial Corp was vulnerable to a simple IDOR (where you change the account number in the URL to get another account), and it evidently resulted in the exposure of hundreds of millions of insurance records that included extremely sensitive information. IDOR is still one of the most common and dangerous vulns a web app can have, and for companies like this they can be devastating. More The US Military is trying to learn how popular movements form and evolve, and to do so they're studying 350 billion social media messages. But it's a Bloomberg article, so maybe they're actually studying bullfrogs for clues about hypertension. More Moody's has downgraded Equifax's rating in some significant part due to its 2017 cyber breach. This is noteworthy because until now, breaches have largely been spackled over in terms of the major financial perspective and at the 6-24 month timescale. This is a positive indication that companies could actually start taking cybersecurity more seriously, and not just at the CISO and IT level, but from the boardroom down. More Advisories: TP-Link Routers Become a Member: https://danielmiessler.com/upgrade See omnystudio.com/listener for privacy information.