Follow us on Google News
Artificial intelligence is changing the speed and scale of cyberattacks, allowing individuals with limited technical experience to attempt operations that previously required specialist knowledge. Cybersecurity experts are warning that the growing availability of powerful AI tools could expose banks, businesses and critical infrastructure to more frequent and sophisticated intrusions.
The concern is no longer limited to whether AI can help hackers write malicious code. Modern AI systems can assist with identifying software weaknesses, testing possible entry points and carrying out multiple stages of an intrusion. This could allow a single attacker to target several organisations in less time than conventional methods would require.
The warning follows a series of security incidents across financial institutions and technology companies. While AI has not been confirmed as the cause of every recent breach, researchers say its growing capabilities are changing the threat environment faster than many organisations can strengthen their defences.
AI Is Lowering The Barrier To Sophisticated Hacking
One example emerged from South Korea, where cyberattacks affected multiple financial institutions. CrowdStrike attributed the activity to a suspected financially motivated individual who allegedly used an AI-powered penetration-testing tool alongside several large language models.
The investigation suggested that AI assistance helped one person target multiple organisations. The case demonstrated how automated tools could reduce the amount of specialist work required to identify weaknesses and pursue unauthorised access.
However, the use of AI does not mean that attacks are operating independently of human hackers. Human operators still select targets, provide instructions and make decisions. The technology can help them perform technical tasks faster and coordinate different stages of an operation.
This distinction matters because the growing threat is not necessarily a sudden arrival of fully autonomous cybercriminals. It is the possibility that existing attackers, including inexperienced individuals, can use AI to increase their capabilities.
Even Advanced AI Companies Face Security Gaps
The risks are also visible inside the technology industry. In a recent security assessment, researchers from cybersecurity start-up Cathedral used AI agents to identify weaknesses in third-party business software connected to Anthropic’s internal systems.
The researchers reportedly gained access to internal business information, including meeting transcripts, salary and recruitment details, and messages discussing advanced AI models. They carried out the work in less than 12 hours through a security testing programme.
Anthropic confirmed that the researchers accessed a limited set of its sales, human resources and information technology systems. The company said they could not access customer conversations with Claude, its responses, source code or model weights. It also said it had disabled the vulnerable access routes and found no evidence that anyone else had exploited them.
The incident highlights a problem that extends beyond AI developers. Organisations often rely on external software for recruitment, communication, administration and customer management. A weakness in one connected service can potentially provide a route into a much larger corporate network.
The episode also shows why security assessments must examine the entire technology environment rather than concentrating only on the main product or application.
Powerful AI Capabilities Are Becoming More Widely Available
Another concern is that advanced cybersecurity capabilities may become increasingly accessible outside a small group of technology companies and approved users.
Some governments and AI developers have restricted access to their most capable models because of concerns about misuse. However, open-weight models, whose underlying model files can be downloaded and run by users, are becoming more capable.
This creates a difficult balance. Restricting access may reduce some opportunities for malicious use, but cybersecurity defenders also need advanced tools to discover vulnerabilities, test systems and respond to attacks.
The same technology can therefore support both sides. AI-assisted security research has helped identify weaknesses that might otherwise have remained undiscovered for years. But once details of a vulnerability become available, attackers may also use automated tools to develop ways of exploiting it.
The challenge is to ensure that organisations have enough time and resources to fix weaknesses before criminals take advantage of them.
Critical Infrastructure Could Face Greater Pressure
The consequences could extend beyond data theft and financial fraud. Energy networks, water facilities, hospitals, telecommunications systems and government services increasingly depend on connected digital infrastructure.
Security experts have called for comprehensive testing of critical systems to identify exploitable weaknesses and provide funding for remediation. They have also stressed the need for stronger cooperation between governments and private companies when responding to incidents.
Earlier warnings from US authorities about remote interference with internet-connected equipment at water facilities demonstrate that weaknesses in operational technology can have consequences beyond the digital environment.
AI is not established as the cause of those incidents. However, the growing availability of automated tools adds another layer of risk to systems that may already have outdated software, limited security resources or weak network controls.
What This Means For Businesses And Ordinary Users
The changing threat landscape means organisations cannot rely only on traditional security checks or respond to vulnerabilities after an attack has begun. They need to maintain an accurate record of their digital systems, install security updates quickly, restrict unnecessary access and regularly test how their networks respond to intrusions.
Third-party software also requires closer scrutiny. Companies should examine which external services can access internal systems and limit the information and permissions available to those services.
For individuals, the risks include more convincing phishing messages, impersonation attempts and fraud based on information obtained through data breaches. Strong, unique passwords, multi-factor authentication and caution when responding to unexpected requests for sensitive information remain important safeguards.
AI could eventually give defenders a substantial advantage by helping them find and repair vulnerabilities before attackers exploit them. But that outcome will depend on investment, responsible deployment and the ability of organisations to act on security findings quickly.
For now, the central warning is that the technical barrier to attempting sophisticated cyberattacks is falling. The organisations responsible for protecting financial systems, public services and critical infrastructure may have less time than before to detect and contain an intrusion.
