FOXTROT/ALFA: Pushback Against Clearview AI, Yahoo Data Breach Settlement, Warcraft 3 Refunds

Welcome to issue 78 of FOXTROT/ALFA for Thursday, 6 February 2020. How you doin' on this fine day? Me? I’m pretty busy. But that doesn’t mean I didn’t take the time to compile the tech news of the day for you. Of course I did!

Google and Venmo Cease & Desist Clearview, Facebook Wants to Watermark Images

Clearview AI is a company that claims it can identify the person on an uploaded picture based on a database of 3 billion images it scraped off the social web. Currently, the app is only available to police in the US, as far as we know. Several social media companies have now sent cease & desist letters to Clearview, saying that this kind of scraping violates their terms of service.

Google, YouTube and Venmo have sent cease-and-desist letters to Clearview AI, a facial recognition app that scrapes images from websites and social media platforms, CBS News has learned. The tech companies join Twitter, which sent a similar letter in January, in trying to block the app from taking pictures from their platforms.

Clearview AI can identify a person by comparing their picture to its database of three billion images from the internet, and the results are 99.6% accurate, CEO Hoan Ton-That told CBS News correspondent Errol Barnett. The app is only available to law enforcement to be used to identify criminals, Ton-That said.

“YouTube’s Terms of Service explicitly forbid collecting data that can be used to identify a person. Clearview has publicly admitted to doing exactly that, and in response we sent them a cease and desist letter,” YouTube Spokesperson Alex Joseph said. “Scraping Venmo is a violation of our terms of service and we actively work to limit and block activity that violates these policies,” Venmo said in a statement. In addition to demanding that Clearview AI stop scraping content from Twitter, the social media platform demanded that the app delete all data already collected from Twitter, according to an excerpt of the cease-and-desist letter.

Meanwhile, Facebook is thinking about invisibly marking images on its platform, so that this kind of abuse can be detected in the future.

Facebook researchers have developed a digital watermarking technique that allows developers to tell if a particular machine-learning model was trained using marked images. “We call this new verification method ‘radioactive’ data because it is analogous to the use of radioactive markers in medicine: drugs such as barium sulphate allow doctors to see certain conditions more clearly on computerized tomography (CT) scans or other X-ray exams,” the eggheads explained on Wednesday.

As far as we can tell, Facebook’s approach is this: you take some photos, watermark them in a way that is invisible to the human eye, label them as normal, and slip them into your image data set. To avoid drawing attention to the watermarking, the marked images should also be labeled as the originals. If an image pre-watermarking was labeled a toaster, it should be labeled a toaster in the data set, too. The idea is not to mess with the labels: just the image data. Now, unbeknownst to you, someone obtains your data set, and uses it to train an image-classification system so that when the AI is shown a photograph in future, it draws upon its training and predicts how a human would label the picture. For example, a photo of a kid on a bike should cause the software to spit out labels like child, bicycle, etc, with varying degrees of confidence: 94 per cent sure it’s a child, 68 per cent sure it’s a bike, two per cent sure it’s a car, and so on.

Later, you come across that model and suspect someone trained it using your data set. When you run your watermarked images through the system, statistical analysis of the neural network’s operation should indicate whether it was trained using those contaminated pictures. This can be done by, for instance, studying the difference between the model’s outputs and what a model trained using vanilla, non-marked images would produce. Alternatively, if you can see inside the model, you could use the network’s weights.

Facebook disclosing this research now is almost certainly motivated by Clearview being exposed publicly by The New York Times last month.

Of course, totally apropos of nothing, Clearview is in hot water at the moment for scraping public photos form web giants for its facial recognition system. Sounds like Silicon Valley doesn’t like having its pictures lifted for other people’s models.

If you want to learn more about Clearview AI, how the company was founded and what exactly it’s doing, may I recommend this podcast episode I recorded yesterday?

Vulnerability in the Windows 10 Driver for Realtek HD Audio

There’s a vulnerability (CVE-2019-19705) in the Realtek HD Audio driver package, which is used with probably the most widespread make of audio cards in consumer PCs, that could lead to an attacker loading malicious code. For the attack to work, the attacker needs administrator privileges, so on its own, it’s is only useful to solidify a hold you already have on the victim’s system.

RAVBg64.exe is an MFC application. MFC provides developers with an option to load localized DLLs (satellite DLLs), which only contain data (resource-only DLLs, which do not contain any code).

There are the missing DLL files which the program tried to load. You can recognize them by the suffix ENU and LOC (e.g. RAVBg64LOC.dll). Despite the fact these are resource-only DLLs, the program loaded our DLL and executed it as a regular DLL.

Realtek used Visual Studio 2005 in order to compile the binary and that is the actual root cause for this vulnerability. It appears that back in 2005, Visual Studio MFC’s implementation was trying to load the satellite-DLL as a regular library with code.

The issue has been fixed in version 8857 of the Realtek HD Audio driver.

Security Hole in WhatsApp’s Desktop Client

WhatsApp’s Windows and Mac clients are at risk of being hacked:

A vulnerability in WhatsApp could be exploited to remotely access a victim’s files on their computer – if they use the desktop client paired with the iPhone app. A patch has been issued and should be installed. Bug-hunter Gal Weizman, from security shop PerimeterX, discovered and reported CVE-2019-18426, a cross-site scripting hole that could potentially allow an attacker to get to the local file system of another user simply by sending a specially crafted message. The security bug was fixed in January by Facebook in WhatsApp Desktop version 0.3.9309 and later.

Please note that the software in question isn’t the browser client (WhatsApp Web).

In short, WhatsApp’s desktop client was built on a version of Electron that used an out-of-date insecure Chromium build, which made it vulnerable to a flaw patched a while back. As a result, users were potentially vulnerable to attack. Users and admins can protect themselves from attack by updating to the latest version of WhatsApp, which is built on a more up-to-date stack.

Half a Million PCs Infected After Owners Downloaded Cracked Software

Don’t download cracked software. Because this is what happens:

Researchers claim more than 500,000 PCs have been left wriggling with malware after a cracked app went on to retrieve further nasties from Bitbucket repos. Security company Cybereason has studied a campaign to deliver “an arsenal of malware” including credential stealers, cryptocurrency miners, ransomware and crypto-coin pinchers.

“It is also able to take pictures using the camera [and] take screenshots,” wrote researchers Lior Rochberger and Assaf Dahan.

Despite the researchers' “Hole in the bucket” headline, the real story here is the risks inherent in users trying to get commercial software for free. Atlassian was quick to remove the malicious repositories reported to them, but the scale of services like this is such that preventing further occurrences is likely to be unrealistic.

Yahoo Data Breach Settlement

If you’re in the US or Israel, and if you’re able to remember your old Yahoo mail address, you might get a few bucks for those huge data breaches a few years back.

Yahoo famously suffered numerous data breaches from 2012 to 2016 (including one in 2013 that affected all 3 billion of its users), and the company is informing users that they can now submit claims for either credit monitoring or cash compensation as part of a proposed class action lawsuit settlement. Yahoo says if you had a Yahoo account anytime between 2012 and 2016 and are a resident of the US or Israel, the class action settlement may affect you. The total amount of the settlement fund is $117.5 million, up from a proposed $50 million settlement that was rejected by a judge.

You must submit your claim online or by mail by July 20th.

But before you get too excited, just remember the Equifax situation; if a ton of people go through this process and opt for the cash payment, you’re not getting anything close to $100.

In addition to the 2013 breach that affected 3 billion Yahoo users, this proposed class action lawsuit also covers a 2014 “state-sponsored” hack that affected more than 500 million accounts, a 2012 “data security intrusion” (though Yahoo says there is no evidence that user credentials, email accounts, or contents of emails were taken from that), and a breach that lasted from 2015 to September 2016 where hackers were able to gain access to approximately 32 million Yahoo email accounts.

Blizzard is Now Offering Refunds for Warcraft III: Reforged

Customer dissatisfaction of Warcraft III: Reforged has gotten so bad, it seems, that Blizzard is now offering refunds for the game:

Blizzard stands by the quality of our products and our services. Normally we set limits for refund availability on a game, based on time since purchase and whether it has been used. However we want to give players the option of a refund if they feel that Warcraft III: Reforged does not provide the experience they wanted. So, we’ve decided to allow refunds upon request for the time being. You may request a refund on our Support Site.

Fallout 76 Coming to Steam

Speaking of bad… Fallout 76 is finally coming to Steam.

On April 7, Bethesda will release its most significant update yet for Fallout 76. The new Wastelanders DLC will be available as a free update for all existing players, bringing many features that Bethesda unveiled during its E3 2019 press conference. Fallout 76 will finally make its way to Steam on April 7 as well. Bethesda announced this last March, but missed its 2019 launch window. The Wastelanders update will be available for new players purchasing the game on Steam, and to those who already own it on PC, PS4, or Xbox One.

Although I have no idea why you would want it, on Steam or anywhere else. It’s a pretty bad game.

No Coffee While You’re Flying That Airbus, Please

I really like coffee, guys. Just as well that I’m not a pilot. Because while I completely understand the reasoning here, I also would be asleep at the controls a lot.

*takes sip of coffee*

The mighty EU Aviation Safety Agency has issued a formal safety directive banning A350 airliner pilots from putting cups of coffee anywhere near sensitive cockpit electronics. It appears that airline pilots simply can’t help themselves when it comes to getting their caffeine fixes. The EU Aviation Safety Agency (EASA) has formally repeated earlier pleas from Airbus not to pour coffee over delicate control panels.

Aviation trade mag Flight Global reported that the cockpit coffee ban came about after two incidents where spilled liquids led to engines shutting down mid-flight “after inconsistent output” from control panels submerged under hot java. The magazine said: “Flight-recorder analysis indicated a high-pressure shut-off valve closure command.”

Yikes. And I thought spilling stuff on my laptop or keyboard was bad…

Also Noteworthy

Some other stories I’ve read today that might be worth a look:


This is an archived issue of my daily newsletter FOXTROT/ALFA. You can find more information about it, including how to subscribe via email, on this page.