The emerging threat of weaponized artificial intelligence

Illustration by Dorathe Victor

Five worthy reads is a regular column on five noteworthy items we discovered while researching trending and timeless topics. In this week’s edition, let’s explore how artificial intelligence and machine learning are weaponized by hackers to fuel cyberattacks.

AI and ML are conquering the world at a rapid pace. AI has made life much easier. In many instances, it speeds up manual processes, reduces costs, and eliminates manual errors. However, it is also known to benefit cybercriminals who employ AI to carry out attacks more effectively.

One example of AI-powered cybercrime is deepfakes. A deepfake, which is a portmanteau of “deep learning” and “fake,” can be used to manipulate faces and voices by combining multiple angles of people’s faces, mimicking their mannerisms and speech patterns, to impersonate individuals. There is so much open source software available online to create deepfakes that it has become almost impossible to identify what is real and what is fake.

ML models require massive amounts of training data to function effectively. The more data fed into the system, the more accurate it gets. However, adversarial ML is a term that defines how an ML system can be tricked into making errors by feeding it incorrect training data, thereby disrupting the ML model. Such data poisoning leads to a vulnerable defense system, allowing attackers to succeed in automated spear phishing attacks by exploiting the weak spots that were classified as positives, which is another threat that is difficult for the untrained eye to detect. 

An AI-powered botnet attack can be extremely dangerous. It doesn’t stop with attacking computer devices—this type of attack is capable of infecting smartphones, surveillance cameras, and IoT devices, too. DDoS is a common form of such attacks; attackers gather as much data as possible to build an ML model, and they use it to train the malware how to predict the system’s defense pattern in order to enable automated decision-making strategies. Such AI-powered attacks are capable of automatically changing their attack strategies based on the defense system’s response.

A sophisticated AI system, when in the hands of a malicious actor, is an imminent threat to security systems; this system can adapt to an organization’s protocols and communication channels by disguising itself and easily spread across multiple devices and networks, bringing down an entire city or a nation’s network in a blink.

The articles below shows how AI and ML are used as an ingredient to create sophisticated malware to perform attacks.

1. Data Poisoning: When Attackers Turn AI and ML Against You

ML and AI are very effective against preventing ransomware; however, cybercriminals can use ML to carry out data poisoning. Adversarial data poisoning can be done two ways: by injecting the wrong training data and destroying the integrity of the system or by generating a backdoor for cyberattacks.

2. Dear enterprise IT: Cybercriminals use AI too

Social engineering attacks like deepfakes and spear phishing attacks have become a growing concern over the past few years. Tailor-made spam and phishing emails target high-profile individuals in an organization using convincing emails and invoices to persuade the user to share confidential data. Weaponized AI can identify weak spots in an organization and adapt to the system by mimicking trusted elements to cause maximum damage.

3. How Criminals Use Artificial Intelligence To Fuel Cyber Attacks

ML requires huge data sets to work effectively and should be stringently monitored and tested to avoid any potential damage it may cause. Threats can occur either through manual errors while designing the system or when someone deliberately wants to weaponize it for attacks by reverse engineering the system to find what data was used to train the system, and then manipulate it. AI-based monitoring and analytics can help identify and alert on any abnormal behavior in the system and combat threats effectively when they emerge.

4. AI-Powered Cyberattacks: Hackers Are Weaponizing Artificial Intelligence

AI is a double-edged sword that can be employed both as a weapon and a solution. Cyberattacks are becoming more frequent, more sophisticated, and harder to detect, leaving organizations increasingly vulnerable to revenue loss, fines, or even bankruptcy. Hackers build AI-enabled malware that can rewrite its code when detected in order to avoid what tripped it up in the past and launch a new attack. Such malware can learn about and adapt to a security system to determine which payload to use to exploit the system on the next attempt.

5. Artificial Intelligence as Security Solution and Weaponization by Hackers 

AI has the capability to adapt to a particular environment or to respond to it intelligently. Hackers develop intelligent malware and execute stealth attacks by learning the patch update life cycle to identify when a system is least protected. AI-based cybersecurity solutions can help organizations by detecting and analyzing unusual activity patterns and locking out users deemed malicious, preventing malware from spreading through the system and accessing system resources.

The powerful nature of AI is being closely scrutinized by organizations and governments. In order to protect data, assets, and intellectual property, every organization has to implement the right AI-based cyberdefenses to ensure they’re properly mitigating the threat of cybercriminals and at the same time preparing to face any forms of threats that might come in the future.