In plain sight: Malicious ads hiding in search results
Sometimes there’s more than just an enticing product offer hiding behind an ad
WeLiveSecurity – Read More
Sometimes there’s more than just an enticing product offer hiding behind an ad
WeLiveSecurity – Read More
The schemes disproportionately victimize senior citizens, as those aged 60 or over were more than three times as likely as younger adults to fall prey to the scams
WeLiveSecurity – Read More
The new school year brings with it new hopes, new subjects, new friends… and new (and not-so-new) video games. After the long summer break, it’s natural for kids to dive back into the cyberworld. When school’s in, there’s less time for hanging out with friends at the mall, so the digital space becomes the preferred meet-up place, including, of course, video games.
But the world of gaming isn’t quite as buddy-buddy as might seem at first glance, so here too cybersecurity is a must. Sure, the games themselves are (mostly) fine — the problem is the parasite scammers and cybercriminals they attract.
Kaspersky experts have dug deep to find out which games and players are most at risk, and what to do about it. See the full version of our report for answers to these, and other related questions.
To fathom the threatscape facing young gamers, our experts analyzed statistics from the global Kaspersky Security Network (KSN). KSN collects huge amounts of anonymous cyberthreat intelligence data that we receive from users on a voluntary basis.
Selecting the most popular kids’ games for the study, we found the top four most-attacked titles from July 2023–July 2024 were Minecraft, Roblox, Among Us and Brawl Stars.
Game name
Number of attack attempts
Minecraft
3,094,057
Roblox
1,649,745
Among Us
945,571
Brawl Stars
309,554
Five Nights at Freddy’s
219,033
Fortnite
165,859
Angry Birds
66,754
The Legend of Zelda
33,774
Toca Life World
28,360
Valorant
28,119
Mario Kart
14,682
Subway Surfers
14,254
Overwatch 2
9,076
Animal Crossing
8,262
Apex Legend
8,133
That’s right, more than three million attack attempts on Minecraft alone! Almost twice more than on second-place Roblox. Why? Because so many players are looking to download mods and cheats for Minecraft, and these often turn out to be malicious apps.
As for the types of threats being spread, the most common are downloaders, adware, Trojans and backdoors. For several years now, malware downloaders have been the most live threat to the gaming industry — downloaders that tout themselves as the “best Minecraft modloader you can get” often turn out to download… backdoors, Trojans and other threats.
While it’s easy to teach your kids to download apps only from trusted sources and use security solutions, keeping them safe from phishing is more of a challenge. Here, it pays to keep your ears and eyes sharp: the more you and your kids know and read about new scams, the better placed you are to spot them. What’s more, most gaming scams tend to follow a pattern.
Pretty much every top kids’ game these days allows (or encourages) players to customize their character with skins that can cost serious money — millions of dollars in some cases! Most kids, of course, don’t have that kind of cash under the bed, so they’re always on the lookout for flashy item giveaways.
One such act of “generosity” was uncovered by our experts. The scammers craftily exploited two things close to young gamer hearts: Valorant and MrBeast. The first is a popular shooter game, while the other is one of the world’s most successful YouTubers, with a 300 million+ subscriber base – mostly kids.
MrBeast and the makers of Valorant probably have no idea about their skin giveaway collaboration on a scam website
The scammers invite gamers to log in to the phishing site using their game account credentials and then to open a treasure chest. Of course, there is no treasure — only a hijacked account.
Most in-game economies are built on two kinds of in-game currency: soft and hard. Soft currency is usually earned through playing the game; hard or premium currency is bought with real-world money. Naturally, it’s the latter that attracts cybercriminals.
For example, one scam asks Pokémon GO players to enter their game account username. That is followed by an “I’m not a bot” verification, after which the player lands on a site promising free in-game currency.
Such calls to action are a ruse to redirect users to a far more serious scam, where not only gaming accounts are at stake, but highly sensitive data like bank details.
“Do such_and_such and win a prize!” is a standard cybercriminal trick. We unearthed such a scam on a Roblox-related phishing site: victims were offered a US$100 Walmart gift card, the same amount for Taco Bell fast food outlets, and, for the especially greedy, US$25,000 in cash. But there’s a catch: first your payment details, please!
Since the youngest gamers don’t yet have payment details of their own, they’ll probably feed their parents’ bank card numbers to the hungry site. And you can only imagine mom and dad’s delight when the next billing statement arrives.
Kids often lack basic cybersecurity skills, so can easily fall into cybercriminal traps for example, when trying to download a free game, a mod or a ‘must-have’ skin. That’s why teaching kids cyber hygiene is one of the most important missions of modern parenting.
Help your child think up a unique strong password, and get them used to using a password manager at an early age.
Tell your child about the risks they might face online.
Our Kaspersky Cybersecurity Alphabet is a fun and informative way to teach your kids about new technologies and basic cyber hygiene, and refresh your own knowledge at the same time.
Install reliable protection for gamers on all devices.
Be in the swim of the latest scams in the gaming world and warn your kids what to watch out for.
Use special apps to keep your kids safe — both online and offline.
For more great security tips for young gamers, check out the full version of our report.
Kaspersky official blog – Read More
Episode 362 of the Kaspersky podcast kicks off with discussion around Brazil’s controversial decision to ban Elon Musk’s X platform. From there the team discuss a story from the BBC around the theft of a voice actors voice, which was used on an A.I platform.
To wrap up the team discuss how scammers are looking to use sextortion tactics in order for you to cough up bitcoin and Apple’s big problem around ‘face swap’ apps and pornography.
If you like what you heard, please consider subscribing.
Top Brazil court upholds ban of Musk’s X
A tech firm stole our voices – then cloned and sold them
Sextortion Scammers Try to Scare People by Sending Photos of Their Homes
Apple’s Huge “Dual Use” Face Swap App Problem Is Not Going Away
Kaspersky official blog – Read More
We regularly hear news about breakthroughs leading to the advent of working quantum computers. For now, such a computer doesn’t exist, so nobody can use one to crack encryption. But when it does arrive, it’ll already be too late to address the problem. That’s why new encryption algorithms that are resistant to both classical hacking methods and quantum-computer attacks are being standardized today. These algorithms are known as post-quantum or quantum-resistant. Support for these algorithms is gradually appearing in everyday devices and applications — they were recently integrated into Google Chrome. This, by the way, immediately exposed compatibility issues within standard organizational IT infrastructures. So, where have post-quantum algorithms already been implemented, and what should IT teams prepare for?
Amazon. The cloud giant introduced a “post-quantum” variant of TLS 1.3 for its AWS Key Management Service (KMS) back in 2020. Since then, the solution has been updated, adapting its configuration settings in line with NIST recommendations.
Apple iOS/iPadOS/macOS. In February 2024, Apple announced an update to the iMessage protocol, which will use the PQ3 quantum-resistant protocol for key exchange. It’s based on the NIST-recommended Kyber algorithm, but also utilizes classical elliptic-curve cryptography, providing dual-layer encryption.
Cloudflare. Since September 2023, Cloudflare has supported post-quantum key agreement algorithms for establishing connections to origin servers (client websites), and is gradually rolling out support for post-quantum cryptography for client connections. The technology is used when establishing a TLS connection with compatible servers/clients, applying a dual key agreement algorithm: classical X25519 for one part of the key, and post-quantum Kyber for the other. This popular combination is known as X25519Kyber768.
Google Chrome. Test support for post-quantum cryptography for establishing TLS connections appeared in August 2023, and as of version 124 in April 2024, it’s enabled by default. The algorithm used is X25519Kyber768.
Mozilla Firefox. Support for X25519Kyber768 for TLS and QUIC appeared at the beginning of 2024, but it’s still not enabled by default and must be activated manually.
Mullvad. This popular VPN service uses the following PQC method: first, a traditional encrypted connection is established, after which a new key agreement is conducted using the Classic McEliece and Kyber algorithms. The connection is then re-established with these keys.
Signal. The messenger implemented the PQDXH protocol in September 2023, using the same X25519Kyber768 mechanism.
Tuta(nota). The popular secure email service allows users to send post-quantum encrypted emails using the X25519Kyber768 algorithm. However, the obvious drawback is that this only works when communicating with other Tuta users.
Although not yet a commercial product, it’s also worth mentioning Google’s implementation of FIDO2 hardware security keys, which use a combination of classical ECDSA and post-quantum Dilithium.
In addition to these, PQC is supported by numerous libraries that serve as the foundation for other products, from email and web servers to operating systems. Notable libraries include OpenSSL and BoringSSL, as well as the experimental branch of Debian. Many of these implementations have been made possible thanks to the Open Quantum Safe initiative, which supports post-quantum forks of popular cryptographic utilities and libraries, available for a variety of popular programming languages.
The algorithms haven’t been sufficiently analyzed. Although the broader scientific community has been conducting cryptanalysis for several years, the mathematical principles behind post-quantum cryptography are more complex. Moreover, experience with classical cryptography shows that serious flaws or new attack methods can sometimes be discovered decades later. It’s almost certain that vulnerabilities will be found in modern PQC algorithms — not just implementation vulnerabilities, but fundamental algorithmic defects.
Key sizes are significantly larger than in RSA and ECC. For example, the Kyber768 post-quantum algorithm has a public key size of 2400 bytes. This leads to a significant increase in data transmission volumes if key renegotiation occurs frequently. In tightly designed or low-power systems, there might not be enough memory for such large keys.
The computational load of PQC is also higher than classical, which slows down operations and increases energy consumption by 2–3 times. However, this issue may be resolved in the future with optimized hardware.
Compatibility issues. All updates to encryption standards and protocols — even classical ones — create complications when some systems have been updated and other related ones haven’t.
Practical issues will primarily affect services using the TLS protocol for connections. TLS is implemented in numerous ways across thousands of products — sometimes with errors. As soon as Google enabled Kyber support by default in Chromium 124, administrators started reporting that Chrome and Edge couldn’t establish connections with web servers, as they would immediately disconnect with an error after the ClientHello TLS handshake. This issue was caused by problem number two: the large key size. As a result, the ClientHello TLS message, which always fitted into a single TCP packet, expanded into multiple packets, and so servers, proxies, and firewalls not prepared for this larger ClientHello message would immediately terminate the connection. Appropriate behavior would involve reading the following packets and agreeing on an older, classical encryption algorithm with the client. A list of incompatible web servers and firewalls affected by this issue is being tracked on a dedicated site, with Cisco notably listed.
If an organization suddenly can’t open any websites, the problem is likely with the proxy or firewall, which needs an update. Until the developers of incompatible applications and devices release patches, a temporary solution is to disable PQC:
using MS Edge and Chrome group policies
in Chrome’s advanced settings: chrome://flags/#enable-tls13-kyber
in Firefox’s settings: about:config -> security.tls.enable_kyber
Administrators are advised to check their websites and web applications by enabling Kyber support in Firefox or Chrome and attempting to access the site. If an SSL/TLS error occurs, the web server needs to be updated.
Standardization is key to preventing a “protocol mess” and compatibility issues. For PQC, this process is ongoing but far from complete.
NIST recently introduced the first full-fledged standards for post-quantum cryptography — FIPS 203, FIPS 204, and FIPS 205. Essentially, these are CRYSTALS-Kyber for key exchange, along with CRYSTALS-Dilithium and SPHINCS+ for various digital-signature scenarios.
European organizations from — ENISA and ETSI to BSI and ANSSI — intend to adopt NIST’s standards but are open to considering additional algorithms if they prove to be better. They all emphasize the necessity of double encryption for critical data — using both post-quantum and classical algorithms simultaneously. Given the novelty of post-quantum algorithms, innovative methods of breaking them may emerge, which is why the second layer of encryption is recommended.
China plans to standardize post-quantum algorithms in 2025. The Chinese Association for Cryptologic Research (CACR) announced the finalists in 2020: Aigis-sig and Aigis-enc (modified relatives of CRYSTALS-Kyber and CRYSTALS-Dilithium) and LAC.PKE.
Meanwhile, the IETF working group responsible for internet protocols will likely endorse the use of cryptography standards proposed by NIST in these protocols.
Kaspersky official blog – Read More
Episode 361 of the Transatlantic Cable podcast kicks off with news around the right to switch off in Australia. From there the team talk about privacy – specifically if you should have to pay to have online privacy.
To wrap up, the team discuss how and why a popular game has attracted so much online attention.
If you liked what you heard, please consider subscribing.
Australians get ‘right to disconnect’ after hours
Should you have to pay for online privacy?
Blockbuster Chinese video game tried to police players – and divided the internet
Kaspersky official blog – Read More
A corrupted hard drive no longer need lead to the loss of all your data. Today there are cloud services: mail is stored in Gmail, files in Dropbox, notes in Apple Notes, and so on. But even with cloud services there’s no doing without backup. Instead of corrupted drives, they present other surprises: for example, they might shutter, hike subscription prices, lose your data, or use it to train AI. And if your internet ever goes down, online-only data is useless.
So as not to be caught off guard by sudden unavailability or policy changes, always back up your data on your own computer and protect it against ransomware. And backups need to be both readable and usable without proprietary software. They should be able to either be exported to common standard formats (PDF, HTML), or migrated to a “backup” app that works offline and without a subscription.
There is no universal recipe here: each online service has its own procedure. Today we look at backing up data in Notion — a knowledge base and note-taking app.
Notion lets you export data in one of three formats: PDF, HTML or Markdown+CSV. You can export a single note, a group of notes, or even an entire database. But only business and enterprise subscribers can do a full export to PDF format.
For most apps, we recommend exporting to HTML format, as it’s free, saves all types of data, and can be viewed in any browser with no special software required.
You can do the exporting on a desktop computer or mobile device. For small amounts of data, a ZIP-archive download is immediate; for large amounts, you receive a download link by email — which arrives with some delay.
To export several notes or a subpage, press the advanced menu icon (•••), select Export, specify HTML as the export format, and include subpages and all types of content (Everything).
An entire workspace can be exported from the desktop app or web interface. Go to the settings, and under Workspace → Settings, click Export all workspace content. In addition to the above settings, be sure to enable Create folder for subpages.
Only workspace administrators have this export option. For teamspaces, the export won’t include other users’ personal (hidden) pages created within the teamspace.
Having unzipped the archive on your computer to a separate folder, you can open the index.html file in it with any browser and freely navigate through your notes.
To not only view saved notes but also be able to edit them without Notion, you have to migrate your data to another, similar app that works offline or on a server under your control. The list of possible alternatives to Notion warrants a long read all of its own, so here we’ll limit ourselves to just two apps that Notion users often recommend as a substitute.
Obsidian is an app for structured data storage that can work entirely offline, free of charge. There’s a paid service — Obsidian Sync — for synchronizing multiple devices, but users manage without it by placing the storage (vault) in an iCloud folder, or by using third-party plugins for synchronization with SFTP, Amazon S3, Dropbox or other services.
To migrate data from Notion to Obsidian:
Perform a full export of the Notion workspace as per the above instructions.
Install Obsidian and the official import plugin.
Create a vault in Obsidian for the migrated data.
Activate the installed plugin under Settings → Community plugins in Obsidian.
Start the import via the button on the vertical command bar on the left.
Select Notion (.zip) as the import file format, and in the dialog, specify the ZIP file downloaded during export.
Enable Save parent pages in subfolders.
Press Import.
Wait for the import to finish.
For very large databases, you may encounter problems with importing embedded ZIP files, in which case see the help page on the Obsidian website.
AFFiNE is an open-source app offering a workspace with fully merged docs, whiteboards and databases, replacing, the developers say, both Notion and Miro. The business model is based on paid plans and AI assistants, but the app can work offline and even function as a standalone server wholly on your own infrastructure.
Content export from Notion is built right into the AFFiNE desktop app, so the procedure is quite straightforward:
Perform a full export from Notion.
Unzip the file to a separate folder on your computer.
Install AFFiNE and create a workspace.
Run the import by going to All pages → New Page → More options → Import page.
Choose import from markdown files, and select the html file from the unpacked folder.
Visit the AFFiNE website for a video guide to importing from Notion.
And remember to protect local backups of your important data against stealers and ransomware with the help of Kaspersky Premium.
Kaspersky official blog – Read More
We’ve conducted a big survey of 10,000 people and have found that many still believe in digital myths. For example, they think that connecting to any public Wi-Fi is totally safe, and that smartphones track all their movements by default.
What’s the reality? Let’s put some of these myths to the test.
Almost half (47%) of respondents think so. At first glance it might indeed seem a tricky task to differentiate between AI and human chats, given that bots have now mastered the Turing test. Nevertheless, it is possible to tell them apart, and we turned to a chatbot to find out how. Come on, AI, tell us how to unmask you!
Chat style. Bots tend to have a more formal or mechanical communication style, while humans may use colloquialisms, jargon and more expressive language.
Response speed. Bots often respond very quickly and consistently, while humans may take a while to respond because they need to think through their response.
Limited topics. Bots may have limited knowledge and fail to understand the context or nuances of a conversation the way that a human does.
Thanks, chatbot! What do you notice about the AI responses? Dry, matter-of-fact, lots of repetition, and minimal deviation from the topic. The information is spot on, and any human expert would say much the same thing… but less like a textbook.
Two-thirds (67%) of respondents are sure their smartphone tracks their geolocation all the time. Well, there’s no law against this opinion. In most cases, such “tracking” is voluntary: users often hand over oodles of permissions to apps, allowing them to feed data to their developers — at least for marketing purposes, but possibly also for real surveillance (albeit unlikely).
And let’s not forget about spyware: nasty little programs that like to masquerade as legitimate apps, but in fact record your calls, read your messages, and sometimes track your movements. Their nastiness is why we advise every single one of those 67% to make sure there’s no spyware on their device. And all that takes is to install reliable protection. Sure, if you’re a bona fide celeb, you can employ the life hack of Sarah Connor from Terminator — she went so far as to carry her phone in a bag of chips to block GPS, Wi-Fi and Bluetooth signals. But chips these days ain’t what they used to be…
Surprised? No less than 28% of respondents turn off their phone or switch it to airplane mode during a face-to-face conversation. Moreover, 26% do this whenever they’re in a public place. Believe us: this anti-spy method is ineffective, and here’s why.
There are Trojans that can record ambient sound — even if the infected smartphone doesn’t have internet access. And as soon as you turn off airplane mode, the malware immediately transfers the collected data to the attackers’ server. The Trojan used in Operation Triangulation, for instance, had this functionality. A far more effective way to guard against cyberespionage is to install a dedicated security solution. And leave airplane mode for flying.
At least 39% of respondents globally think so — respondents, it seems, who didn’t read our study of public hotspots in Paris. In July, on the eve of the Olympics, we analyzed the security of around 25,000 unique Wi-Fi hotspots in the French capital, and found a full quarter of them to be wholly insecure — many protected by outdated security protocols. The results apply to any city in the world — things are hardly different in, say, Moscow, Berlin, Tokyo or Sao Paulo.
So avoid connecting to public Wi-Fi unless if you absolutely must, in which case:
Don’t buy anything online.
Don’t log in to personal accounts without two-factor authentication.
Enable a reliable VPN on your devices, as well as a firewall on your laptop.
Disable file-sharing and AirDrop on your devices.
Four out of ten people concur. We’re among the other six out of ten who understand that incognito mode won’t make your surfing totally private. Sure, it doesn’t save your browsing history, doesn’t remember information entered on visited sites, and doesn’t store data in the browser cache — in other words, incognito mode leaves no traces of browsing on your device. But it doesn’t hide your IP address, so someone could get a fix on your location if desired. It’s also possible to expose your identity if you’re logged in to a site.
Incognito browsing is perfect for when you want to leave minimal traces on your device. For example, when searching for gifts for family — especially if you all use the same computer and browser. This way, the browser won’t remember your actions or spoil the surprise with untimely contextual ads. For more tips on what else you should and shouldn’t do in incognito mode, see our separate post.
One of the missions of Kaspersky — of our Kaspersky Daily blog in particular — is technological evangelism. We tell you all about new technologies and threats, uncover the most sophisticated scams, and do everything we can to shatter digital myths and superstitions wherever they may be.
But the contents of this post are just the tip of the digital iceberg; our original report is bursting with even more eye-popping myths. Help us bust them all by sharing this post with family and friends — especially if they’re just starting out building a cybersecure future for themselves.
Kaspersky official blog – Read More
After many years of research and testing, in mid-August 2023, the U.S. National Institute of Standards and Technology (NIST) finally introduced fully-fledged post-quantum encryption standards — FIPS 203, FIPS 204, and FIPS 205. So let’s discuss them and see why they should be adopted as soon as possible.
First, let’s briefly outline the threat quantum computers pose to cryptography. The issue lies in the fact that quantum computing can be used to break asymmetric encryption. Why is this important? As a rule, today’s communication encryption typically uses a dual system:
All messages are encrypted using a symmetric algorithm (like AES), which involves a single key shared by all participants. Symmetric algorithms work well and fast, but there’s a problem: the key must be somehow securely transmitted between interlocutors without being intercepted.
That’s why asymmetric encryption is used to transmit this key (like RSA or ECDH). Here, each participant has a pair of keys — a private and a public one — which are mathematically related. Messages are encrypted with the public key, and decrypted only with the private one. Asymmetric encryption is slower, so it’s impractical to use it for all messages.
The privacy of correspondence is ensured by the fact that calculating a private key from the corresponding public key is an extremely resource-intensive task — potentially taking decades, centuries, or even millions of years to solve. That is — if we’re using traditional computers.
Quantum computers significantly speed up such calculations. Specifically, Shor’s quantum algorithm can crack private keys for asymmetrical encryption much faster than its creators expected — in minutes or hours rather than years and centuries.
Once the private key for asymmetric encryption has been calculated, the symmetric key used to encrypt the main correspondence can also be obtained. Thus, the entire conversation can be read.
In addition to communication protocols, this also puts digital signatures at risk. In the majority of cases, digital signatures rely on the same asymmetric encryption algorithms (RSA, ECDSA) that are vulnerable to attacks by quantum computers.
Today’s symmetric encryption algorithms, on the other hand, are much less at risk from quantum computers than asymmetric ones. For example, in the case of AES, finding a 256-bit key using Grover’s quantum algorithm is like finding a 128-bit key on a regular computer. The same applies to hashing algorithms.
The primary task for cryptographers has become the development of quantum-resistant asymmetric encryption algorithms, which could be used in key transfer and digital signature mechanisms. The result of this effort: the post-quantum encryption standards FIPS 203, FIPS 204, and FIPS 205, introduced by the U.S. National Institute of Standards and Technology (NIST).
FIPS 203 describes a key encapsulation mechanism based on lattice theory — ML-KEM (Module-Lattice-Based Key-Encapsulation Mechanism). This asymmetric cryptographic system — which is resistant to quantum algorithm attacks — is designed to transfer encryption keys between interlocutors.
ML-KEM was developed as part of CRYSTALS (Cryptographic Suite for Algebraic Lattices) and is also known as CRYSTALS-Kyber, or simply Kyber.
FIPS 203 features three parameter variants for ML-KEM:
ML-KEM-512: Security level 1 (equivalent to AES-128);
ML-KEM-768: Security level 3 (equivalent to AES-192);
ML-KEM-1024: Security level 5 (equivalent to AES-256).
FIPS 204 defines a digital signature mechanism, also based on algebraic lattices, called ML-DSA (Module-Lattice-Based Digital Signature Algorithm). Previously known as CRYSTALS-Dilithium, this mechanism was developed within the same CRYSTALS project as Kyber.
FIPS 204 has three parameter variants for ML-DSA:
ML-DSA-44: Security level 2 (equivalent to SHA3-256);
ML-DSA-65: Security level 3;
ML-DSA-87: Security level 5.
The third standard, FIPS 205, describes an alternative digital signature mechanism: SLH-DSA (Stateless Hash-Based Digital Signature Algorithm). Unlike the other two cryptosystems, which are based on algebraic lattices, SLH-DSA is based on hashing. This mechanism is also known as SPHINCS+.
This standard involves the use of both the SHA2 hash function with a fixed output length, as well as the SHAKE function with an arbitrary length. For each base cryptographic-strength level, SLH-DSA offers sets of parameters optimized for a higher speed (f — fast), or a smaller signature size (s — small). Thus, FIPS 205 has more variety — with as many as 12 parameter options:
SLH-DSA-SHA2-128s, SLH-DSA-SHAKE-128s, SLH-DSA-SHA2-128f, SLH-DSA-SHAKE-128f: Security level 1;
SLH-DSA-SHA2-192s, SLH-DSA-SHAKE-192s, SLH-DSA-SHA2-192f, SLH-DSA-SHAKE-192f: Security level 3;
SLH-DSA-SHA2-256s, SLH-DSA-SHAKE-256s, SLH-DSA-SHA2-256f, SLH-DSA-SHAKE-256f: Security level 5.
For now, the threat of quantum algorithms breaking asymmetric encryption is mostly theoretical. Existing quantum computers lack the power to actually do it in practice.
Until last year, it was believed that sufficiently powerful quantum systems were still a decade away. However, a 2023 paper suggested ways to optimize hacking using a combination of classic and quantum computing. As a result, the timeline for achieving quantum supremacy seems to have shifted: RSA-2048 could very well be broken within a few years.
It’s also important to remember the concept of HNDL — “harvest now, decrypt later” (or SNDL — “store now, decrypt later”). Attackers with significant resources could already be collecting and storing data that can’t currently be decrypted. Once quantum computers with sufficient power become available, they’ll immediately begin retroactive decryption. Of course, when this fateful moment comes, it will already be too late, so quantum-resistant encryption standards should be implemented right now.
The ideal approach to deploying post-quantum cryptography based on established IT industry practices is hybrid encryption; that is, encrypting data in two layers: first with a classical algorithm, then with a post-quantum one. This forces attackers to contend with both cryptosystems — significantly lowering the chances of a successful breach. This approach is already being used by Signal, Apple, Google, and Zoom.
Kaspersky official blog – Read More
Thanks to scientists at the University of the Republic (Uruguay), we now have a much better understanding of how to reconstruct an image from spurious radio emissions from monitors; more specifically — from signals leaked during data transmission via HDMI connectors and cables. Using state-of-the-art machine-learning algorithms, the Uruguayan researchers demonstrated how to use such radio noise to reconstruct text displayed on an external monitor.
Sure, it’s not the first attempt at a side-channel attack aimed at reconstructing an image from radio signal emissions. A method of intercepting radio noise from a display in a neighboring room — known as a certain TEMPEST attack — was described in a study published in… 1985! Back then, Dutch researcher Wim van Eck demonstrated that it’s possible to intercept a signal from a nearby monitor. In our post about the related EM Eye attack, we talked extensively about these historical studies, so we won’t repeat ourselves here.
However, van Eck’s experiment has lost much of its usefulness today. It used a monitor from 40 years ago with a cathode-ray tube and analog data transmission. Also, the captured image back then was easy to analyze, with white letters on a black background and no graphics. Today, with a digital HDMI interface, it’s much more difficult to intercept the image, and, more importantly, to restore data. But that’s precisely what the Uruguayan team has managed to do.
Data is transmitted digitally to the monitor via an HDMI cable. The volume of data involved is vast. The computer transmits 60 or more frames to the monitor every second, with each frame containing millions of different-colored dots. Using a software-defined radio (SDR), we can intercept signals generated by this data stream. But can we then extract useful information from this extremely weak noise?
Schematic of the new spying method proposed by the Uruguayan team. Source
The authors called this attack Deep-TEMPEST — a nod to the use of deep-learning AI. The diagram clearly shows how noisy the intercepted data is before processing: we see a discolored shadow of the original image, in which only the location of the main elements can be guessed (a browser window with an open Wikipedia page was used for the experiment). It’s just about possible to distinguish the navigation menu at the top and the image in the center of the screen, but absolutely impossible to read the text or make out the image.
Image captured and processed by Deep-TEMPEST. Source
And here’s the result after processing. The picture quality hasn’t improved, so making out the image is no easier. But the text was recognized in its entirety, and even if the machine-learning algorithm tripped up on a couple of letters, it doesn’t greatly affect the final result. Let’s look at another example:
Deep-TEMPEST attack result in detail. Source
Above is the captured image. Some letters are distinguishable, but the text is basically unreadable. Below is the original image – a screenshot fragment. In the middle is the image after processing by the machine-learning algorithm. Some adjacent letters are hard to discern, but overall the text is quite easy to read.
The Uruguayan team’s main achievement is that they developed their own method of data analysis. This was partly due to enhanced neural network training, which allowed text recognition from a rough image. To do this, the team needed pairs that consisted of an original screenshot and the corresponding SDR-captured image. Building a dataset big enough for training (several thousands of pairs) is a difficult, time-consuming task. So the researchers took a slightly different path: about half of the dataset they obtained by displaying an image on the screen and intercepting the signal; the other half they simply generated using a self-written algorithm that gives a reliable picture of the captured information based on the relevant screenshot. This proved sufficient to train the machine-learning algorithm.
The team’s second stroke of genius was the use of a neural network that delivered high-quality results without much expense. The test bed was created from relatively affordable radio-data interception tools; open-source software was used. As we said, HDMI carries vast amounts of data to the connected monitor. To analyze spurious radio emissions during such transmission, it’s important to intercept a large spectrum of radio frequencies — the bigger the band, the better the result. Ideally, what’s needed is a high-end SDR receiver capable of capturing a frequency band of up to 3200 megahertz — a piece of kit that costs about US$25 000. In this case, however, the researchers got by with a USRP 200-mini receiver (US$1500) — capable of analyzing a much narrower frequency band of up to 56 megahertz. But thanks to the enhanced neural network trained to recognize such partial information, they could compensate for the lack of raw data.
Deep-TEMPEST attack test bed. On the left is the target computer connected to a monitor. Key: (1) antenna, (2) radio signal filters and amplifier, (3) SDR receiver, (4) laptop for intercepting radio emissions and analyzing the data. Source
Open-source software and libraries were used to process the data. Code, screenshots and other resources have been made available on GitHub, so anyone who wishes to can reproduce the results.
In the 1999 novel Cryptonomicon by Neal Stephenson, one of the characters, upon discovering that he’s being monitored by “van Eck phreaking”, starts making things difficult for those spying in him by changing the color of letters and replacing the monochrome text background with a video clip. Generally speaking, the countermeasures against TEMPEST-type attacks described by Stephenson a quarter century ago are still effective. You can add noise to an image such that the user won’t even notice — and interception is impossible.
Naturally, the question arises: is the juice worth the squeeze? Is it really necessary to defend against such highly specialized attacks? Of course, in the vast majority of practical cases, there’s nothing to fear from this attack – much better to focus on guarding against real threats posed by malware. But if you work with super-valuable data that super-professionals are after, then it might be worth considering such attacks as part of your threat model.
Also, don’t disregard this study out of hand just because it describes interception from an external monitor. Okay, you might use a laptop, but the image is sent to the built-in display using roughly the same principles — only the transmission interface may be slightly different, while the radiation level will be slightly lower. But this can be addressed by refining the algorithms and upgrading the test equipment. So hats off to the Uruguayan researchers — for showing us once again just how complex the real world is beyond “software” and “operating systems”.
Kaspersky official blog – Read More