Cyber ​​threat forecast for 2025: AI and targeted attacks

The world of malware (malicious program) continues to develop at a dizzying pace, with technological changes occurring at an increasing pace and the geopolitical tensions in the world creating a reality where cyber threats become more complex and dangerous than ever before. Against this background, CYFOX’s research division issues a forecast detailing the expected cyber threats for 2025.

“The year 2025 is expected to be a turning point in the cyber field, with an increasing focus of attackers on high-value targets, in combination with the increasing use of artificial intelligence and other advanced technologies to create custom made malware,” says Nir Yehoshua, director of research at CYFOX. In addition, trends such as the wider use of cloud technologies, the proliferation of devices connected to the Internet (IoT), and increasing dependence on digital services, will increase the threats to organizations and individuals alike.

Moving from broad attacks to more targeted attacks

While in the past we have seen many broad attacks designed to hit as many victims as possible, in 2025 attackers are expected to focus on specific high-value targets. This trend will be aided by the increasing use of artificial intelligence, which allows identifying unique weaknesses in specific systems and organizations quickly and efficiently.

In addition, the ease of searching on LinkedIn and social networks allows attackers to collect personal information about potential victims, which enhances the ability to carry out more targeted and precise attacks.

For example, during last October, a malware called ESET Wiper was discovered that was part of a targeted cyber attack aimed at Israeli companies that are customers of ESET Israel. Malicious ESET Wiper is attributed to a pro-Palestinian attacker known as WIRTE, and it is designed to use data destruction and anti-analysis mechanisms to delete data from disk drives, thus damaging the functioning of critical systems and disrupting corporate activities.

CYFOX estimates that attackers will continue to use artificial intelligence to upgrade their attack methodologies and tools. In 2025, we expect to see an increasing use of AI to create custom made malware, which can bypass traditional security systems, such as EDR, Firewall, IDS, learn the defense mechanisms of the systems, and accordingly change their operating patterns.

Dataset-based attacks are a type of cyber-attacks that utilize training data (datasets) of artificial intelligence (AI) or machine learning (ML) systems to influence the performance of the system or penetrate it. They are especially common in systems that depend on learning data to make autonomous decisions or predictions.

Attacks of this type may become a major threat in 2025. The fear is that attackers will try to exploit the dependence of machine learning systems on high-quality training data and change or pollute the datasets. These actions cause the models to learn incorrect operating patterns or develop weaknesses that may be used by the attackers for targeted attacks or to deceive the system.

In recent years and especially since the Iron Swords War, Israel has consistently faced state cyber attacks carried out by hostile countries such as Iran or attack groups supported by them. These attacks target critical infrastructures such as energy systems, water, transportation, strategic targets, government bodies and defense industries, with the aim of destabilizing, harming the economy and disrupting the functioning of the state.

The increasing geopolitical tensions in the Middle East are expected to lead to a further increase in cyber attacks of this type and Israel is expected to face attacks by military or civilian attack groups that will attempt to disrupt critical infrastructures and to gain access to sensitive information. The tension in the region poses a significant challenge not only to Israel, but also to its partners in the West, who may serve as indirect targets, through supply chain attacks.