How to protect yourself against AI data breaches

A Kaspersky report has recommended seven ways to safeguard against Artificial Intelligence (AI) data breaches. The cybersecurity company noted that with the state of AI today, there is a need to stay alert to avoid uncertainties.

Here are 7 ways to safeguard against data breaches:

Learn to spot deepfakes

Always double-check unusual requests while contacting people through another channel. The rapid advancement of neural networks has allowed scammers to move from deepfake videos of celebrities to inexpensive and relatively massive attacks on specific individuals using fake voices and images of anyone.

Deepfakes were first used to promote financial pyramids or fake charities, but now targeted scams are in play, for example, calls from the victim’s boss or a loved one. It’s now easier than ever to make a video of someone you know well asking for money or something else.

Block your calendar for monthly backups

It is now more important than ever to schedule backups daily, weekly, or monthly, depending on the data type.

Kaspersky recommends that backup must be two-way: backup data on your phone and computer to cloud storage and download cloud data to local storage. An example of the former is photos on your phone. An example of the latter is Gmail messages.

This way, you will be protected against many problems: computer crashes, smartphone theft, ransomware attacks, house fires, your favourite recipe site being shut down, movies and music disappearing from streaming platforms, sudden hikes in cloud-hosting charges, and so on.

Read also: The first significant mass-scale artificial intelligence (AI) deception case

Use passkeys rather than passwords

Global tech companies are beginning to replace passwords with passkeys. With it, biometric verification or PIN codes will be used to log in to sites rather than passwords.

The check is carried out locally on your computer or smartphone, after which the device decrypts from its storage a unique cryptographic key for the website in question, which recognises you by this key. This method is more reliable than a combination of a password and a one-time code and is easier and faster to use.

Change all old passwords

Although passkeys are being used now, passwords are still very prevalent. Old passwords created years ago with little thought to length or strength can be brute-forced without too much trouble.

To thwart password crackers, go through all your passwords and, reset any that are short (fewer than 12 characters) or very old, and create new ones in accordance with best security practices. As you know, passwords should never be reused, so it’s best to generate new ones and store them in a reliable password manager.

Switch to private messages

According to Kaspersky, moving to messaging platforms that offer end-to-end encryption is necessary.

For example, the arrest of Pavel Durov, Telegram’s founder, raised fears that intelligence agencies could start snooping on users’ correspondence.

Enter your card details less often

So as not to be left guessing if your payment data is safe, and not to have to contact banks to reissue cards after every major leak, save your card to a reputable, secure service and use it to pay for purchases wherever possible, (for both offline and online purchases.) This will make it harder for attackers to intercept your payment data and reduce the likelihood of damage in the event of a large store or online service hack.

Use AI assistants securely

Given that AI is now literally in the palm of your hand, offering, at times, quite personal advice, it is essential to understand the rules for safe chatbot use to keep yourself and others out of harm’s way.

Double-check AI advice, especially when asking for information about medicines, investments, or other queries where errors are costly. Chatbots are known to hallucinate, so never blindly follow their tips.

Disable AI features unless you know what they are for. The smart craze is driving companies to integrate AI even where it’s not needed. Disable AI if you’re not actively using it.

Never give personal information to AI. Photos of documents, passport details, and financial and medical information are rarely needed for AI to function correctly. Given that such data may get stored for a long time and used for AI training and thus be more likely to leak, it is better not to upload such data in the first place.

Don’t chat with family and friends through AI. Such automation is rarely helpful and won’t help maintain closeness.