Artificial intelligence is rapidly influencing our lives. As we become more and more dependent on technology, everyday gadgets are becoming smarter by the day.

The Jetson’s lifestyle no longer sounds farfetched. Virtual assistants are now smart enough to schedule our daily lives, and smart appliances make controlling our homes easier than ever.

With AI now going mainstream, big companies are coming up with new prospects to capitalize on AI technology.

But there’s a catch. Despite revolutionizing the way we live, AI technology and its vulnerabilities make users a prime target for hackers to exploit.

In this blog post, we’ll take a look at how hackers are actually fooling smart gadgets to snoop up on our privacy.

AI Taking Over

Compared to a decade ago, smart gadgets have exponentially grown in popularity. Gone are the days when you had to walk across the hall just to flick a switch. Now you can simply summon up your voice assistant and control everything with voice commands.

If we take a look at statistical data from 2010 to 2019, the market value of smart appliances has increased from $40 million to 26.149 million U.S dollars in 2019.

Statista-trend-of-smart-home-appliances

The overwhelming popularity of smart appliances is also evident from the comprehensive study published online which analyzed the conversations of millions of smart home technology consumers.

According to the collected data, consumers are largely optimistic about smart home technology with 60% positive sentiments.

Smart Home Sentiment Trend – Cepro

Since more and more consumers are interested in adopting smart home technology, overall discussion about smart homes has risen in searches from 25,000 in 2012 to over 250,000 in 2017.

Based on overwhelming popularity, tech giants like Google, Apple, and Tesla are integrating neural processing custom AI hardware to power their smartphones and self-driving cars.

But most notably, LG has been pushing AI integration in almost all of its ThinQ products for the past few years. Now the company is even planning to add custom hardware to power future robot vacuum cleaners, refrigerators, washing machines, and even air conditioners.

LG’s Neural Engines are so powerful, they can mimic human brains to improve the processing of deep learning algorithms even without a network connection.

These AI Chips can recognize locations, objects, chemical changes, and intelligently distinguish voice characteristics from the environment.

This brings us to why there’s a good chunk of people out there who believe AI-powered home appliances can be the reason behind their compromised privacy.

After all, news about smart devices being hacked, or Alexa devices creeping in on the conversation is becoming quite common.

Open-Source AI Technology & Its Vulnerabilities

“The oldest and strongest emotion of mankind is fear, and the oldest and strongest kind of fear is fear of the unknown”
H. P. Lovecraft

According to chief evangelist of Carbonite Norman Guadagno, “Hackers are no longer solely focused on PCs and the encryption of files”.

“They are now targeting new platforms, like IoT-connected devices that serve as entry points, to obtain valuable information.”

Since AI and machine learning technology is much more accessible thanks to open source applications, hackers can potentially exploit vulnerabilities to take full control.

Attackers can carry out denial-of-service attacks and control-flow hijacking attacks that can lead to compromised systems or glitched deep learning applications.

So on that note, let’s check out a few vulnerabilities that existed in one of the most popular open-source AI development applications – TensorFlow.

1) TensorFlow

Created by Google, TensorFlow is one of the most widely used open-source machine learning frameworks that’s used by giants like eBay, Twitter, and even Uber.

Similar to other powerful machine learning applications, TensorFlow is built on a layered framework with 3rd party packages such as NumPy and LibRosa to aid matrix multiplication, input parsing, and GPU optimizations.

The layered model brings TensorFlow close to 887K lines of code and increases the risks of vulnerabilities caused by added complexities.

Despite being backed by Google and capable of developing neural networks with full-fledged computational models, TensorFlow is known to have a few vulnerabilities. Just check out this massive vulnerabilities list.

For instance, before the security patch was released, TensorFlow suffered from DoS software bugs that caused infinite loops, abrupt crashes, and even exhausted memory.

More specifically, the now patched vulnerability existed in the NumPy python package, which is an integral part of the TensorFlow framework.

TensorFlow NumPy python package vulnerability

Not only that, but TensorFlow also suffered from a buffer overflow vulnerability that existed in the application’s OpenCV library.

Although the OpenCV vulnerability is now been patched, this type of bug is usually associated with “hijack control flow”. If exploited, attackers can remotely compromise targeted systems deep learning applications.

These type of bugs are not just limited to TensorFlow, but also exists in other machine learning applications like Torch, Caffe, PyTorch, and others.

Eventualities of How Smart Appliances Can Be Hacked

We all know the recent Alexa scandal that involved Amazon employees listening in on voice recordings captured through Echo.

But recently enough, the company has also been grilled for pitching facial recognition technology to law enforcement agencies and retailers to gauge human emotion such as fear, sadness, anger, surprise, disgusted, calm, and confused in real-time.

Even Apple has recently come forward and apologized for allowing contractors to grade Siri users based on stored voice recordings. Shockingly enough, Siri captured illegal acts, confidential information, and even sexual interactions of users.

Considering these recent cases, one can imagine what hackers can do if they gain access to smart appliances like refrigerators, surveillance cameras, printer, and even cars.

Smart Refrigerators

If you’re forgetful, you’ll likely love smart refrigerators that can remind you to get rid of expired products.

That being said, since smart fridges are connected to your Wi-Fi network and have access to your username and password, a hacker can potentially breach your privacy.

If you’re not accustomed to using secure networks, a hacker can use your stolen credentials to even access your banking services.

Smart Security Cameras

Smart surveillance cameras are perfect for catching a thief in the act, but they’re also a gateway for hackers to spy on you.

One way that hackers use to tap into security cameras is through a technique called “credential stuffing.” By using stolen credentials made available by hackers from other data breaches, hackers can take complete control of your surveillance system.

What makes this type of attack so easy is the wide availability of stolen credentials online. In fact, in recent years, hackers have publically released credentials of over 8.2 billion accounts.

Smart Printers

Printers might seem harmless, but when workplace printers get hacked, it becomes a recipe for disaster.

For instance, a hacker by the name of Andrew Auernheimer hacked printers inside dozens of college campuses to distribute neo-Nazi propaganda.

In another similar case, hackers took control of company printers and distributed stolen credentials of the company’s email service provider account.

By exploiting default printer credentials, or router ports like 721–731, 9100, and 515, a hacker can take full control of your printer.

Countermeasures

By incorporating smart appliances like lights, surveillance cameras, thermostats, and door locks, we’re inviting hackers and digital thieves in our homes.

We’ve all heard about the case of Laura Lyons whose Nest security camera was hacked and used for spying and broadcasting false reports of a nuclear missile attack.

This begs the question, is there a way we can make our smart appliances and gadgets more secure?

1) Secure Your Network

This might seem like a no brainer, but making your home network secure should be your top priority. But with millions of stolen credentials available online, you’ll need more than just a password to protect your network.

Fortunately, there are tools like Virtual Private Networks that can establish a secure network connection between your device and the secure VPN server. This way, all of your network traffic gets end to end encrypted, securing you from would-be hackers.

2) Choose Powerful Passwords

Believe it or not, most people make the mistake of using the same passwords for multiple accounts. Even if one account gets hacked, attackers can take over all associated accounts.

Change your device’s default username as well. Most people make the mistake of leaving their devices’ usernames and passwords as it came from the factory. This makes it easy for hackers to brute force their way into your home surveillance system.

Now granted that remembering passwords for 10 different accounts can be near impossible, there are apps that can help you generate and remember powerful password combinations.

3) Check device logs

Check the logs on your smart home appliance. Most smart home security cameras can show the IP addresses of devices that access your camera. If you see anything suspicious, change your password and notify the concerned authorities.

4) Pay attention to software updates

Update as often as possible. Once you buy a security product for home protection, pay close attention to the software updates required. If you update frequently, you’ll be able to further strengthen your line of defense against online threats.

Wrapping Up!

Without a doubt, AI is the feature. Over the coming years, AI will become more powerful and prevalent than ever. Future smart appliances and gadgets will become even smarter and hopefully much more secure.

Companies like LG are promising to deliver a better artificially intelligent future with better built-in security and protection against hackers

We’ve come so far ahead in time that we can’t take AI out of our lives. But what we can do is make it much secure for future generations.

Image Credit: Gerd Altmann/Pixabay

Source: https://datafloq.com/read/open-source-ai-powered-appliances-that-can-be-remotely-hacked/8516