One of the most important technical developments of the twenty-first century, artificial intelligence (AI), is transforming businesses, economies, and people's daily lives. AI has the potential to fundamentally change the way we work, communicate, and solve complicated problems because of its capacity to process enormous volumes of data, recognize patterns, and make defensible conclusions. Moreover, it has the capability to automate routine operations, freeing up human resources to concentrate on more strategic and innovative projects. Artificial intelligence (AI)-driven solutions improve productivity, cut down on errors, and streamline processes, from chatbots that answer customer questions to predictive analytics that optimize supply chains. Humans are making the most of the benefits provided by intelligent equipment. However, what if we tell you that according to the latest research, AI data theft is a possibility in 2023? Can’t believe what you read? Read ahead to learn how AI is capable of stealing your sensitive data
Can AI steal your information?
Recently, researchers at Cornell University, found out that AI tools are capable of stealing your information, through keystrokes. It further explained that an attack can be conducted by an AI leading it to steal passwords by listening to the rhythm of your typing. This information was found in an experiment where scientists trained an AI model on the sound of keystrokes and placed it near a phone. The built-in microphone listened attentively to the keystrokes on a Macbook Pro and reproduced with up to 97% perfection. The team repeated the experiment during a Zoom call in which the keystrokes were recorded with the laptop’s built-in microphone. In this test, AI was surprisingly 93% accurate in replicating the keystrokes. Even when the same test was applied on a Skype call, the results were 91% accurate. Before you decide to throw your keyboard into the garbage, you should be aware that the volume of the keyboard was not exactly the culprit for sensitive information disclosure. In fact, the reality is that AI was trained on the waveform, intensity, and time of each keystroke in order to recognize them. For example, you can press a key a second later compared to others, due to your different typing style. However, the attack would take the form of malware installed on your phone or any device that has a microphone. It just requires you to collect data from your keystrokes and put it in an AI model by listening on your microphone. According to the information found, the researchers used a CoATNet an AI image classifier for the attack, and developed the model on 36 keystrokes on MacBook Pro Pressed 25 times each.
How to be Safe from AI Data Theft?
With your safety at risk, here is what you can do to keep your data protected. First of all, it is wise to avoid typing your password through features such as Windows Hello and Touch ID. You can also rely on a good trustworthy password manager, which eliminates the threat of exposing your passwords but also lets you use random passwords for your accounts. If you are thinking buying a new keyboard can save you from AI stealing information, then you are wrong! Even the best keywords can be vulnerable to attack, so it won’t make a difference. Just a few weeks ago this attack was detected by the latest AI tools such as ChatGPT. FBI has warned earlier about the cons of the latest OpenAI tool and how it is suspected to be used in launching criminal campaigns. With such tools, security researchers have detected new challenges like adaptive malware too.
Here are some other tips to protect yourself from AI Data Theft:
Watch what you post online: Even if you don't think the information you provide online is sensitive, AI systems can nevertheless learn a lot about you from it. Use secure passwords for all of your internet accounts and exercise caution while posting on social media.
Use encoding: Make sure your sensitive data is encrypted before transferring it over the internet. It will be less likely to be intercepted by AI systems or other bad actors thanks to this.
Utilize a VPN: Your traffic can be made anonymous with a VPN, which makes it more difficult for AI systems to monitor you. If you use generative AI tools or other services that gather personal data, this can be useful.\
Recognize the dangers: It's critical to be aware of the risks associated with using AI systems, especially when dealing with sensitive data. Before using the system, be sure you comprehend how it functions and what data it gathers.
AI Might Outsmart Humans Very Soon
It has been all about laughs since the new AI tools has come into the market. They have proved to be essential for humans, as their extraordinary intelligence has made daily tasks for us super easy. Needless to say, due to them we are able to finish our work on time. So far, it has been a fun ride but with challenges like data theft coming up, is a bit eerie. It can be a sign of a problem rising, that we thought we never had to face. However, for now, only future can tell, if AI will outsmart humans or not but for now, it is our best friend… well, sort of!