Machine learning

What is machine learning?

This model learns and improves based on data and experience. A typical example of machine learning is generative artificial intelligence (GAI). GAI can generate text, video, audio, and images. Different GAI services specialize in various forms of content they generate.

Examples of GAI

ChatGPT (Text incl. code), Microsoft Copilot (Text incl. code), DALL-E (Images), MidJourney (Images), Bard (Text incl. code), Google Gemini (Text incl. code), Udio (Music), Boomy (Music), Google DeepMind’s V2A (Video and audio incl. dialogue).

In development, we have Genie (under Google’s DeepMind’s V2A) which can generate simple video games. The development of GAI is progressing at a rapid pace.

Can GAI be a cyber threat?

Ja, GAI kan være en trussel for både bedrifter og privatpersoner.

Automated attacks

Cybercriminals can use GAI to automate and customize attacks. This means that criminals can use GAI to learn while it attacks a target. For example, GAI can be used to find and exploit vulnerabilities faster than a human can, and unlike humans, it will not overlook vulnerabilities due to fatigue or inattention.

Some are already speculating that quantum computers and artificial intelligence will be combined to attack targets in the future.

Phishing and social engineering

GAI can be used to generate emails and messages that are very convincing and difficult to distinguish from genuine communication. Mass phishing emails are already widespread, but with the use of GAI, scammers can send out mass phishing emails without the typical spelling errors found in such emails. This way, they can trick even more people into giving away personal information and sensitive data.

Generation of malware

GAI can be used to generate new and more sophisticated malware that can more easily bypass traditional security systems. Traditional antivirus and firewalls will therefore be at a disadvantage when facing artificial intelligence.

Manipulation of data

GAI can be used to manipulate the dataset used to train other AIs. They can sneak in misinformation, thereby manipulating how artificial intelligence makes decisions and is trained. This can lead to incorrect, and in the worst case, dangerous outcomes.

Deepfakes

Above, language models like Google’s DeepMind V2A were mentioned. Deepfakes can be used to generate fake videos of people doing things that put them in a bad light. For example, it could make Norway’s Prime Minister say something that could damage our relationship with Sweden or Denmark.

Can GAI be used in cyber defense?

Yes, GAI can be used in cyber defense with the following methods:

Threat detection

GAI can analyze large amounts of data to identify patterns and anomalous and unusual activity. With the help of GAI, a consultant can detect cyber threats more quickly.

Automated response

GAI can automatically detect and respond to cyber attacks by isolating affected resources and initiating necessary measures.

Enhanced monitoring

GAI can continuously monitor network traffic and user behavior. Additionally, GAI can oversee large amounts of data simultaneously and keep track continuously.

Using more than GAI in cyber defense

In addition to using GAI in these AI-driven cyber defense solutions, they also use automated decision-making, anomaly detection, risk assessment, and more.

Sicra and machine learning

Several of Sicra’s partners provide security solutions that implement artificial intelligence, including machine learning in their solutions. If you want to know more, we recommend the main article on Artificial Intelligence.

Our experts are also ready to advise on the use of artificial intelligence. We can assist your company technically, and our CISO-for-Hire service can help you become NIS2 compliant.