A wave of severe disruptions to digital systems and services has raised concerns about cybersecurity globally. In just six days, major problems affected Apple, Microsoft, JPMorgan Chase, the US Federal Aviation Administration (FAA) and even the internal network of the US National Security Agency (NSA). Although the incidents have not been officially linked to each other and in some cases have been attributed to technical flaws, experts warn that artificial intelligence could fundamentally change the way cyberattacks are carried out.

The outages have reportedly affected essential services, from communications and electronic payments to air traffic control systems. At the same time, warnings from the FBI and Google have raised alarm about the use of autonomous artificial intelligence tools by hacking groups.

The problems began on November 12, when a widespread outage affected Microsoft's Outlook, Teams and Exchange Online. The outage lasted for about 24 hours and the company attributed it to a previous software issue, combined with reduced server capacity.

On November 14, the FAA's air traffic control systems suffered a serious outage, according to the report. An unusual data surge affected the aging control equipment, knocking out radar at 17 of the 22 control centers. The situation reportedly led to widespread flight cancellations across the United States.

JPMorgan Chase also faced a trading service outage for several hours, with the Downdetector platform recording thousands of reports from users. The bank attributed the problem to a software bug and denied any malicious activity.

Meanwhile, on November 17, Apple users were hit by severe problems. Many reported difficulties making payments, downloading apps, and logging into iCloud.

That same day, sources cited by the New Republic and CNN said NSA Net, the top-secret internal network through which US intelligence officials exchange classified information, was also down. The agency reportedly delayed the arrival of some employees while an initial assessment of the situation was conducted, but there was no immediate indication of a cyberattack.

An unusual explanation for these developments was given by US Homeland Security Secretary Markwayne Mullin during an interview on Fox News. He said that a source in the technology industry had raised the possibility that the outages were related to the rapid development of artificial intelligence and programs that can write and correct code themselves.

According to Mullin, these systems could interfere with existing infrastructures faster than they were designed to handle. However, he also mentioned that his source was 98% certain that this was not a cyberattack.

Researcher Gary Marcus has expressed concerns about the uncontrolled use of artificial intelligence agents and has called for their development to be temporarily restricted.

The risk is not only related to the possibility of systems experiencing defects, but also to the use of artificial intelligence by hackers to automate attacks.

According to the material, one of the tools mentioned is ARTEX, a multi-agent framework published on the GitHub platform, which can be connected to various artificial intelligence models, including Claude, GPT, DeepSeek, and Grok.

The Chinese group Storm-0940 allegedly used this tool in mid-October to target the supply chain of the company ConnectWise. The attackers attempted to penetrate more than 2,250 Microsoft 365 customer environments, forcing the company to suspend several critical services.

According to the FBI, in some cases, using these tools can take attackers from identifying a target to gaining privileged access in less than three hours.

Google is also monitoring the development of these technologies through a report from its threat intelligence team, GTIG. Researchers have identified a system called *exploit_agent* that can analyze application code before and after security updates are released, identify closed vulnerabilities, and automatically create tools to exploit them on systems that haven't been updated.

John Hultquist from GTIG warns that hackers don't need to invent new methods, as they are using known vulnerabilities, but at a much greater speed and scale than before.

Cybersecurity expert Apostolos Kritikopoulos, in an analysis for Newsbomb, also links the future risk to advances in mathematics and quantum computers.

He notes that in 2026, artificial intelligence has helped solve mathematical problems that had remained open for decades. Among the examples is a counterexample found by an internal OpenAI model to a conjecture proposed by mathematician Paul Erdős in 1946.

According to the expert, this is of particular importance for digital security, as many of the encryption systems used today rely on mathematical problems that are practically impossible to solve by traditional computers.

Quantum computers could change this balance, necessitating the creation of new encryption methods that will withstand their computational power.

The article cites various estimates of the number of qubits needed to break 2048-bit RSA encryption, as well as a study that predicts the possibility of developing such a computer within the next ten years. However, current machines are still far from this capacity.

The consequences of cyberattacks are not limited to technology companies. A ransomware attack can lock down a business's systems, steal data, and paralyze activity for days or weeks.

An example is the attack on Jaguar Land Rover in 2025, which disrupted production at British factories for five weeks. The damage to the UK economy was estimated at £1.9 billion.

The risk also extends to banks, energy systems, and internet-connected devices. In 2015, a cyberattack on Ukrainian energy companies left about 225 customers without electricity. Meanwhile, researchers at Princeton University have demonstrated through simulations that controlling a large number of internet-connected household devices can create serious problems for the power grid.

According to expert Kritikopoulos, such a scenario could become even more dangerous if thousands or millions of artificial intelligence agents are coordinated to undertake different parts of the same attack.

He argues that the answer cannot simply be to stop the development of artificial intelligence, as the technological race between the great powers has already advanced. According to him, priority should be given to the creation of new encryption systems, personnel training, tested crisis management plans and offline backups.

The main warning is that artificial intelligence is giving attackers new speed and capacity, while critical infrastructure in many countries continues to rely on legacy systems. For this reason, cybersecurity is increasingly becoming an issue that affects not only companies and institutions, but also the daily lives of citizens.

 

© BalkansWeb
To become part of the group "Balkanweb" just click: Join Group and your request will be approved immediately. Groups Balkanweb